studio.el changed, so the committed build input went stale the moment the merge
landed. CI compiles dist/soul.c, not the .el files. The stamp gate named
studio.el and refused; this is the regeneration it asked for.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
el_cgi_init loaded the declared identity into runtime globals at startup and
printed it. Nothing read it back out — no accessor existed and it writes no
state, so every consumer still read identity from the mutable state store.
studio.el's dharma_registry read state_get("soul_principal"), a key with no
producer anywhere in the tree, and reported an empty principal under a heading
reading 'Principal Covenant v1'.
Adds cgi_name / cgi_dharma_id / cgi_principal / cgi_network / cgi_engram and
points dharma_registry at the compiled constant.
Read-only on purpose. There is deliberately no setter: publishing these into the
state store would have been one line, passed the same test, and recreated exactly
the runtime-mutable copy IDPROTO claims 1-2 forbid.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Installing the fixed compiler exposed a blind spot in this gate. On clean main,
with no source changed, soulc-stamp reported OK while the committed amalgam had
gone stale by a line — because the fingerprint covered .el sources and not the
toolchain that turns them into dist/soul.c. That is exactly the class of silent
divergence the gate was written to close, and it had it.
The stamp now fingerprints the elc binary alongside the sources. Demonstrated: with
the old stamp the gate passed after a compiler swap; with this change the same
condition fails, naming __compiler__.
dist/soul.c regenerated under the installed compiler (1,205,027 bytes) and verified:
builds from its own committed input, the declared principal is present in the
resulting binary, interface 110 routes in / 110 out.
Differential evidence that the new compiler is a strict superset — same sources,
both compilers:
neuron soul 1 differing line, the el_cgi_init emission
engram server.el 0 differing lines
mcp-wrapper 0 differing lines
mcp-proxy 0 differing lines
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
el_cgi_init loaded the declared identity into runtime globals and printed it, and
nothing read it back out. No accessor existed and it writes no state, so every
consumer still read identity from the mutable state store. studio.el's registry
read state_get("soul_principal") — a key with no producer anywhere — and reported
an empty principal under a heading reading 'Principal Covenant v1'.
Adds cgi_name/cgi_dharma_id/cgi_principal/cgi_network/cgi_engram. READ-ONLY on
purpose: there is deliberately no setter. Publishing these into the state store
would have been one line and would have recreated exactly the runtime-mutable copy
IDPROTO claims 1-2 forbid ('not modifiable by any runtime mechanism including
environment variables, configuration files, or API calls').
dharma_registry now reads the compiled constant. cgi_id keeps its state read
deliberately — the runtime instance id is a different fact from the compiled
dharma_id, and conflating them would hide a binary running under an id its own
declaration never claimed.
Measured, same corpus, binary the only variable:
deployed engine -> "principal":""
accessor build -> "principal":"william-christopher-anderson"
interface: 110 routes in, 110 out, nothing removed
Requires the codegen fix in el (fix/cgi-identity-emission); without it the
declaration is never compiled in and the accessors return empty.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Five sites in chat.el splice state_get("soul_identity") into the system prompt.
Nothing has written that key since b163fa6 deleted the producer days after 601e0fe
added it on 2026-05-02. Every chat turn since built its prompt with an empty
identity section, and nothing reported it.
Found by the #132 state-key gate within an hour of that gate being rebased onto
main — a read with no producer treated as a build error rather than a silence.
Restored verbatim rather than repointed. soul_identity is an env-configurable
persona line; soul_identity_context is the graph-derived DNA/values/memory-philosophy
block. Aiming the five reads at the latter would have substituted different content
and called it a repair. Whether the chat prompt should also carry that block is a
separate question, left open rather than smuggled in.
Verified by the gate that found it: dead reads 6 -> 1, with the chat.el baseline
entry now reported STALE. Rung: BUILT and gate-verified; not end-to-end chat-verified.
Closes#137. Refs #132.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Five sites in chat.el read state_get("soul_identity") and splice the result into
the system prompt, beside the voice, security and capability rules:
chat.el:737, 1745, 2620, 3425, 3480
Nothing has written that key since b163fa6. The producer was added 2026-05-02 in
601e0fe and deleted by the awareness refactor days later. Every chat turn since has
built its system prompt with an EMPTY identity section, and nothing reported it.
Found by the #132 state-key gate, which treats a read with no producer as a build
error rather than a silence. That is the entire argument for that gate.
RESTORED VERBATIM, NOT IMPROVED. soul_identity is an env-configurable persona LINE.
It is not soul_identity_context — the graph-derived
[INTELLECTUAL-DNA]/[VALUES]/[MEMORY-PHILOSOPHY] block written at soul.el:184.
Repointing the five reads at that block would have substituted different content and
called it a repair. Whether the chat prompt should ALSO carry the graph-derived block
is a real question and a separate one; it is not smuggled in here.
Verified by the gate that found it: dead reads 6 -> 1, and it now reports the
chat.el baseline entry as STALE — 'entries that no longer match anything; delete
them'. The remaining one is studio.el's soul_principal, untouched by this change.
Rung: BUILT, gate-verified, boots. NOT end-to-end chat-verified — proving the
prompt now carries the line needs a live provider call, which I have not run.
Refs #137, #132
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The runtime-vs-owner divergence check. Its absence let a ~24,000-node loss run for
weeks with every boot reporting green, which is most of why the last two days were
spent rediscovering by hand what this route would have said.
Follows the spec rather than inventing a metric: CGI provisional
05-detailed-description.md Stage 1 calls for an annotated characterization of the
graph's structure, so the route returns findings with score null BY DESIGN. A number
here would be a fabrication dressed as rigour.
Rebased across 27 commits of drift. The rebase merged cleanly at source level and
that was misleading — dist/soul.c held one side's code and not the other, because
git resolved the amalgam as an ordinary file. The stamp gate from 9fd8c11 caught it
on its first real use. Without it this would have landed an engine containing the
audit but none of the 08-09 engine work, or the reverse: neuron#133 again.
Verified before merging, not after:
stamp OK dist/soul.c matches the sources (1,204,442 bytes, 1,259 bodies)
builds from committed input 920,776 bytes
interface 108 -> 110 routes, nothing removed
the route answers stage 1, annotated_characterization, findings present
Closes#142. Refs #91.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The rebase merged cleanly at source level but left dist/soul.c holding one side's
code and not the other — main's regenerated amalgam vs this branch's. The stamp
gate added in 9fd8c11 caught it on its first real use:
FAIL: dist/soul.c is STALE. It does not match the current .el sources.
Sources that changed: neuron-api.el, routes.el
Without that gate this branch would have looked clean and shipped an engine
containing the structural audit but none of the 08-09 engine work, or the reverse.
That is exactly neuron#133, which once hid five merged fixes including a P0.
Regenerated from the rebased sources: 1,204,442 bytes, 1,259 inlined bodies.
Verified after: stamp OK; builds from its own committed input (920,776 bytes);
interface 108 -> 110 routes with nothing removed, adding /api/neuron/audit/structural;
and the route answers — stage 1, assessment_kind 'annotated_characterization',
score null by design, with findings.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`runStructuralAudit` has been an advertised MCP tool with nothing behind it: the
dispatcher GET'd /session/begin and returned that unrelated session digest under
an audit tool's name. Meanwhile the failure the audit would have caught ran
silently for about three weeks — the soul reporting 103,089 nodes while the
engram, which OWNS persistence, held ~79,900, a crash discarding the difference,
and every boot reporting green throughout, because nothing in the system ever
compared the two sides.
WHAT THE PATENT SPECIFIES, AND HOW IT SHAPED THIS
CGI provisional, 05-detailed-description.md, "Stage 1: Structural audit 430".
Two clauses did the design work. First the four things the module evaluates:
the density and typed distribution of causal edges; value/execution-record
consistency; the richness and connectivity of the self-model; and wonder-
manifest authenticity. Second, and decisively: it "produces a coherence
assessment 432 — NOT A BINARY SCORE but an annotated characterization of the
graph's structural properties."
So every finding carries its numbers AND a plain-language note saying what
they mean and how they were obtained. There is no pass/fail and no composite
health figure, and `"score":null` is emitted explicitly so a reader cannot
mistake its absence for an omission.
WHAT IS IN STAGE 1 (four findings)
owner_runtime_divergence — the motivating case. Runtime counts vs the owner's
own GET /api/stats, the delta, and the trend against the previous audit, so
a second call answers "is the gap growing?" rather than restating it.
self_model_connectivity — the three identity pillars plus the self root:
present, content length, one-hop degree. This RETIRES the Claude-side vitals
identity block, which lived outside the system it was checking and went on
reporting green while the memory-philosophy pillar was absent from the live
graph. Asking the running soul is the designed mechanism; a shell probe was
the fourth patch on the same hole.
typed_edge_distribution — exact counts against the claim-10 vocabulary, plus
density, plus a separate count of LOWERCASE near-misses ("causes" vs
"Causes"): "the vocabulary is unused" and "the vocabulary is misspelled by
the write paths" are different defects with different fixes.
orphans_and_dangling_edges — the tool's own long-standing promise.
WHAT IS DEFERRED, AND WHY IT IS DATA RATHER THAN A COMMENT
Value/execution-record consistency and wonder-manifest authenticity ship as a
`deferred` array that MEASURES the populations they would need (Prediction and
WonderQuestion nodes) and reports those counts as the reason. Both are ~0 today
— WonderQuestion because of a known write/read node-type mismatch. Asserting
value coherence or a pull-weight correlation on an empty population would be a
fabricated result, which is worse than a stated gap.
MEASUREMENT HONESTY: EXACT WHERE CHEAP, SAMPLED WHERE NOT, ALWAYS LABELLED
Counts, edge typing and self-model connectivity are exact. Orphan and dangling
rates are sampled, because engram_find_node_index is a linear scan — an
exhaustive dangling check is O(nodes x edges), ~2.2e9 string compares at today's
scale. Samples are UNIFORM across the whole population (str_index_of_all gives
every edge offset in one pass, so any index is O(1); json_array_get would have
been O(n^2)), never head-of-list, and each figure ships with its own sampled /
population / exhaustive fields. ?edge_sample= and ?node_sample= at population
size run either check exhaustively. The real fix is an id index in the runtime.
ONE BUG THIS FOUND IN ITSELF, CAUGHT IN TEST
http_get does not return "" when the owner is unreachable — it returns a JSON
error object. Testing only for "" made a DEAD owner read as reachable with
node_count 0, so the audit reported 100% divergence and named it data loss.
Reachability is now proved by the presence of the node_count field, and the
owner's raw reply is attached. A confident wrong answer is exactly what this
route exists to stop.
Edge findings need relation labels and the runtime has no edge-enumeration
builtin, so they use the same scratch export GET /api/graph/edges already uses
(engram_save to TMPDIR, never the owner's canonical file — #117). That is a
large write on a large graph, so this is a manual route, not a timer; ?edges=0
skips it.
neuron-api.el:900-1273 handler + helpers
routes.el:567,752 GET and POST /api/neuron/audit/structural
mcp-wrapper/src/main.el:113,682 tool description + dispatch off /session/begin
dist/soul.c regenerated (1255 bodies)
Rung: E2E-VERIFIED. Soul built from this branch (gen-soul-amalgam + cc-brain,
921,192 bytes, 16 warnings, 0 errors), booted on throwaway ports 7893/7896/7897
with throwaway HOMEs against a stub owner on 7894. Three scenarios pass: owner
reachable (runtime 62 vs owner 42, delta 20 / 32.2%, trend flat on the second
call; 12/20 edges claim-10 typed, 3 lowercase near-misses; 50/62 orphans, 3/20
dangling — every figure matches the fixture by construction), owner unreachable
(reported as a finding with the raw reply, not a crash), and file mode (owner
"none", divergence undefined). Reached end-to-end through the MCP tool via a
locally built wrapper. verify-soul-contract.sh: GATE PASS, 27/27 routes +
immutability. No process left running; live :7770 and :8742 untouched (GET only).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Consolidates the 2026-08-09 engine work. Each piece was measured in an isolated lab
against the real corpus, gated, and verified on the operator machine before landing.
WHAT IT CONTAINS
Semantic retrieval + the #146 delete fix, shipped together on purpose. The
semantic leg was once wired into engram_search_json, which seven sites use as a
keyed read and then delete every result from; one runs at every boot. Measured
cost when live: ~234 real records destroyed per startup, identity among them.
Retrieval must never reach production without the engram_recall_json split.
Self-seeded compiled context. The design: 'Every compilation query begins at the
self-model node and traverses outward.' Ours seeded from a hardcoded string, so
compiled context held 0 identity records. Now 18, bounded the same way every other
list in that handler is — the unbounded version is what set self_neighbors to []
after it closed the socket on every call.
Connections on write. CCR claim 29 requires linking candidates with typed edges
'rather than appending as unlinked content'. Unlinked append was the only
behaviour we had: 5% of nodes connected, no edge created by any write in 19 days.
dist/soul.c drift becomes a build failure (#133), and deploys now build from that
same committed input rather than a scratch amalgam — one lineage, provenance
recorded, enforced by a gate in the deployer.
MEASURED, all on the real corpus or the live machine
rephrased-question recall 0/43 -> 20/43 (bar was >=6 queries; nonsense controls held 10/10)
identity in compiled context 0 -> 18
connections per memory 0 -> ~2 on write; 2,828 backfilled into history
no data loss across boots 44,418 / 44,418 / 44,418 over three restarts
FOUR DEFECTS FOUND BY MEASURING RATHER THAN READING, each fixed here
an inert test arm (both arms byte-identical but for a float rounding artifact);
an association hook wired into a path the request never takes (zero edges across
four writes); an identity exclusion matching lowercase 'self' that let a memory
link into the identity graph, whose verification shared the blind spot and printed
PASS; a telemetry filter matching hyphenated 'state-event' that missed
'INTERNAL STATE EVENT'. Common shape: a filter tested a proxy for the property it
cared about. Assert the property.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Until now deploys were built by build-soul.sh from a scratch amalgam while CI
compiled the committed dist/soul.c. Two lineages — 'what runs' and 'what the repo
says builds' were different artifacts. That is #133/#111 in another costume, and it
is how a round-9.1 brain shipped matching no committed source at all.
build-soul-from-dist.sh asserts dist/soul.c matches the .el sources, compiles it
with CI's own flags (-O2 -DHAVE_CURL -rdynamic; the CI comment explains -rdynamic —
without it the runtime cannot resolve its HTTP handler by name and the binary serves
nothing on every route), and writes a .provenance sidecar recording the dist/soul.c
hash, the stamp hash and the commit.
deploy_binary.sh gains GATE 0: refuse any soul without provenance, or built from a
dist/soul.c other than the one in the repo now. The override
(NEURON_DEPLOY_UNSTAMPED=i-accept-two-lineages) exists deliberately — a gate with no
escape hatch gets bypassed by disabling the gate, which is worse than one that
announces itself.
Verified before enforcing, so the sanctioned path is not a broken one:
interface parity with the deployed binary 108 routes in, 108 out
boots and serves 2s
carries today's work 24 self-neighbours, 17 identity records
GATE 0 refuses an unstamped binary exit 10
GATE 0 accepts the stamped one, deployed durability PASS
Production now runs 4845db3d, built from dist/soul.c at 9fd8c11. After the swap:
identity in context 17, connections still form on write, semantic recall intact,
mind and store at delta 0.
Refs #133, #111
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
#133 regenerated the amalgam once and said so itself: 'Nothing in the tree
regenerates this file. Only a human running the recipe. It lags in batches, never
per-change, and it will drift again.'
It drifted again. Every binary deployed on 2026-08-09 was built by build-soul.sh
from a scratch amalgam that never touches dist/soul.c, so the committed build
input fell 2,761 bytes behind the sources by a different route than #133 describes.
Auto-regeneration is not available: the CI workflow records that elc needs 24GB+
of virtual memory and would OOM the runner. So the build cannot regenerate the
file. It can refuse to compile a stale one, for free and with no compiler.
tools/soulc-stamp.sh records a content fingerprint of every .el source at the
moment the amalgam is generated. --check recomputes and compares; divergence exits
1 and names the changed files and the recipe. Wired into CI ahead of the compile.
dist/soul.c regenerated from current sources: 1,176,361 -> 1,179,122 bytes, 1,247
inlined bodies (gate wants >=1200), and verified to compile clean at 903,552 bytes.
Demonstrated to FAIL on the bad input, per postmortem 0004's rule that a gate which
only passes on good input proves nothing:
fresh stamp -> OK, exit 0
one .el modified -> FAIL, exit 1, names memory.el
source restored -> OK, exit 0
Refs #133, #111
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The design rejects what this system did: promotion must link candidates 'using
typed semantic edges rather than appending as unlinked content' (CCR claim 29).
Unlinked append was the only behaviour we had. Measured 2026-08-09: 14,214 edges
over 80,936 nodes, 5% of nodes connected to anything, and no edge created by any
write since 2026-07-19 across 27,000+ new nodes. A memory with no connections is
unreachable by spreading activation, so retrieval degrades to literal matching.
On write, a memory is now linked to its top related existing memories.
Bounds, each bought with a specific failure:
- max 3 edges per memory (link_memories.py's cap: precision over spray)
- never link to identity. The existing policy is explicit that 'memories must
not pollute the self traversal by similarity; only an explicit citation may
touch identity'.
- never link telemetry (state-event, soul-response, boot_count, loop-outcome,
search-result): ~97% of daily write volume. Linking it would add thousands of
noise edges a day and re-flatten the graph in the name of connecting it.
- fail-soft: a failed association never fails the write
- associate only AFTER durability, so no edge points at a node that did not
persist — that is the dangling-edge defect the 08-09 cleanup removed 830 of
Two defects found by measuring rather than reading, both fixed here:
1. Hooking mem_store alone produced ZERO edges across four real writes. The HTTP
memory route writes via wt_node directly; mem_store serves only the awareness
telemetry paths we refuse to link.
2. A lowercase-only identity check let a memory link to 'Self — Values
(grounded)' — the exact pollution the policy forbids. My verification shared
the blind spot and printed PASS. Now uses str_lower.
Measured, lab, same corpus: ~1-2 edges per memory; identity edges unchanged at
475; zero identity leaks post-fix including an adversarial batch of six memories
written about values. Edges route through wt_edge so they reach the owner.
Rung: E2E-VERIFIED in an isolated lab. Not deployed.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The design is explicit — 'Every compilation query begins at the self-model node
and traverses outward... structural reachability from the self-model node is a
precondition for any node to appear in compiled context.' (will-anderson
patents/drafts/engram-claims.md, Self-Seeded Activation. DRAFT, not a filed
provisional — cited as such.)
Measured before: compiled context held 0-1 identity records out of 10, because
compilation seeds from a hardcoded text string and never from the self. Even an
explicit 'my values identity who I am' query returned a boot counter and
state-events.
This restores the designed behaviour without repeating the failure that set
self_neighbors to [] originally: that was an unbounded ~90KB neighbour dump which
closed the socket on every call. Same bound as every other list in the handler.
Cap is 24 rather than 8 because the self root's first 8 neighbours are tag nodes
('neuron', 'tier:note', 'disposition:experimental', 'imprint', 'traversal') that
crowd out the substantive identity records behind them. The root has 34
neighbours of which 23 are identity.
Measured after, same corpus, binary as the only variable:
identity records in compiled context 0 -> 17
response size 10,806 -> 18,985 bytes
interface gate 108 routes in, 108 out, none removed
Rung: E2E-VERIFIED in an isolated lab. Not deployed — this is the identity layer.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The accumulated retrieval stack (iterations 1-9) put the claim-24 semantic
leg and the claim-10 associative leg on engram_search_json — the function
~40 internal .el call sites already used as a KEYED read. Seven of those
sites delete every record that comes back ("prune all existing X nodes,
keep exactly one"): memory.el:176, sessions.el:250/268/444/523,
soul.el:359.
mem_boot_count_inc() calls engram_search_json("soul:boot_count", 50) and
engram_forget()s all 50 results. With a lexical leg that returned 1 record.
With a semantic leg it returns 50 — the 49 nearest neighbours of the STRING
"soul:boot_count" — and the soul deletes them.
MEASURED on the harness corpus, isolated, read-only, zero writes from any
caller: 234 node records destroyed in a single boot. The deletion list is
the soul's own lookup result list, in rank order. Casualties include 6
Knowledge nodes, a layer-1 "CORE IDENTITY - GENESIS, LINEAGE" Memory, the
value node kn-58874a74, and the gold answers to 8 of the 75 gold-set
queries. After the fix: 1 deletion, which is the one the code intends.
THE BOUNDARY, from Will. Claim 24 authorises the vector index "to respond
to EMBEDDING SEARCH QUERIES by returning the node records whose embedding
vectors have the highest cosine similarity to a query vector". A keyed
state read is not an embedding search query; it is the identifier-keyed
retrieval of claim 23 ("node records are stored under a key encoding the
node identifier"). One function served both, so a nearest neighbour of
"soul:boot_count" was treated as a boot counter.
So: engram_search_json returns to its lexical contract, and the legs move
to engram_recall_json, which is what /api/neuron/recall reaches — the route
the MCP wrapper, the app, and this harness all call. Retrieval quality on
that route is unchanged by construction.
MEASURED, 75-query extended gold set, embedded corpus, vs the iteration-9
baseline: +3 / -0 (q15, q28, q60), p=0.2500, hit@5 53.8 -> 58.5%, latency
1.02x, every regression guard held, nonsense 10/10. Net +3 against a floor
of 6 is NOT-SHOWN and I am not calling it an improvement. The deliverable
is the defect.
Diagnostics kept, env-gated (EG_DIAG / EG_DIAG_ID), zero cost when unset:
node/embedding census at load, per-query leg dump, and a FORGET log — the
last is the regression detector for exactly this class of bug.
LIMIT, stated: handle_api_search_knowledge still uses the lexical function.
It is a retrieval surface and arguably wants the legs, but nothing in this
harness measures it, so I did not change unmeasured behaviour.
Measured on the 75-query extended gold set (iteration 8's held-out extension)
against the certified stack baseline results-stack-ext.json, on the embedded
corpus. Three runs of the candidate, zero drift.
A. CLAIM 24 WITHOUT THE THRESHOLD - net +4, NOT-SHOWN, kept in the tree.
fixed : q14, q25 (in-sample paraphrase), q43, q52, q63, q67 (held-out)
broken : q15 (paraphrase), q28 (associative)
8 discordant, McNemar exact p = 0.2891, floor is 6.
heldout_paraphrase 16.7% -> 30.0%, paraphrase 61.5% -> 69.2%.
Every regression guard held: exact_rare 6/6, phrase 7/7, nonsense 10/10,
superseded 2/3. Latency FLAT: p50 641 -> 632 ms.
The in-sample half (+q14 +q25 -q15 -q28 = 0) was already on record in
iteration 7's cmp-nogate.json, so only the held-out +4 is new.
B. ASYMMETRIC TASK PREFIXES ON THE EMBEDDER - net -5, REVERTED in this commit.
Rationale was sound and the prediction was wrong, which is why it was worth
measuring: nomic-embed-text is an asymmetric retrieval encoder and this file
embedded query and document bare on both sides. Prefixing does exactly what
the model card implies for the far-away cases - it rescued q42 (gold at
GLOBAL COSINE RANK 25,564) and q39 - but it re-ranks the whole space and
broke more than it fixed:
fixed : q24, q39, q42
broken : q18, q19, q22, q31, q43, q44, q52, q63
heldout_paraphrase 30.0% -> 23.3%, paraphrase 69.2% -> 53.8%.
The corpus and the reproducer are kept (embed-corpus-prefixed.py,
snapshot-pre-repair-20260806-embedded-prefixed.json) so nobody re-runs it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Two changes to the ONE leg that generalises. Iteration 8 measured that out of
sample the semantic leg contributes 100% of the stack's gain and the graph leg
contributes nothing, so this is where the remaining headroom is.
1. THE 0.60 FLOOR IS A PER-QUERY LOTTERY, AND CLAIM 24 HAS NO THRESHOLD IN IT.
06-claims.md l.148: "respond to embedding search queries by returning the
node records whose embedding vectors have the HIGHEST COSINE SIMILARITY to a
query vector, independently of the spreading activation traversal." A
ranking. ENGRAM_EMBED_SEED_MIN is defined at el_runtime.c l.6094 as the
HippoRAG seed-JOIN threshold and l.6102 admits the read-path leg merely
"reuses" it. Measured on the 30 held-out paraphrases: the query's own top-1
cosine ranges 0.564-0.680, so the constant keeps a rank-1 answer for one
query and discards a rank-1 answer for the next. Six golds sit at global
cosine rank 1-2 scoring 0.564-0.589 - discarded by nothing but the constant.
What holds the nonsense controls is the corpus-vocabulary gate (nhits == 0),
not this floor. Cosine clamped to [0,1] per 05-detailed-description l.69.
2. THE VECTORS THEMSELVES ANSWER THE WRONG QUESTION. EL_EMBED_MODEL defaults to
nomic-embed-text, an ASYMMETRIC retrieval encoder trained with task prefixes.
Embedding query and document bare - as this file did on both sides - measures
topical similarity rather than answer-hood. eg_embed_fetch now takes the task
prefix: EL_EMBED_QUERY_PREFIX on the three query call sites, EL_EMBED_DOC_PREFIX
on the two backfill sites. Restores no claim, and says so: Will specifies only
"computed by an embedding model over the node's content" (l.17), so the model
is his and its correct use is ours. It is the substrate under claim 24 -
the index is only as good as the vectors in it.
Reproducer for the derived corpus: tools/retrieval-eval/embed-corpus-prefixed.py.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The retrieval match test is a raw substring scan, so a query token matches
anywhere INSIDE a corpus word: "throom" matches "bathroom". Measured over the
38-query gold set on this corpus that is not a rare accident - q28's lexical
leg is 36,954 records of which only 13 contain a query token at a word start
(99.96% mid-word noise), six further queries carry ~20,500 mid-word-only
records each, and the nonsense control q35 returns 7 records ALL of which
match only mid-word.
istr_contains_wordstart() anchors a token to a word start (preceding char not
alphanumeric) while still matching suffixes, so "value" still hits "values".
That empties the lexical leg for gibberish, and the nhits==0 corpus-vocabulary
gate (iteration 6's mechanism, feat/claim24-unfloored-semantic) then makes the
whole query decline rather than let the semantic leg answer it.
Measured vs feat/bm25-lexical-leg on the embedded corpus, 2 runs each,
0 queries of run-to-run drift on both sides:
net +1 (nonsense:q35), 0 losses, McNemar p=1.0 -> NOT-SHOWN (floor is 6)
nonsense clean 2/3 -> 3/3; exact_rare 100%, phrase 100%, paraphrase 61.5%,
associative 66.7%, superseded 2/3 all UNCHANGED
latency p50 1184 -> 543 ms (0.46x)
Iteration 6 called q35 "a DEFECTIVE CONTROL ... cannot be cleaned without
breaking the lexical leg". It can: the defect was the match primitive, and
cleaning it cost nothing.
Also committed: results-wsclaim24.json + cmp-nogate.json, a measured negative
for bundling the claim-24 unfloored semantic leg on top (gains q14/q25, breaks
q15/q28/q33/q34, net -2) - it independently reproduces iteration 6's q15/q28
losses and shows unflooring REQUIRES the vocabulary gate.
Reproducers: legs.py (leg-level replica, reproduces baseline hit@5 exactly on
all 38 queries), policy2.py, ceiling.py, wb2.py.
hit@5 68.6% -> 74.3%, phrase 71.4% -> 100%, MRR@10 0.461 -> 0.502, latency
p50 0.97x. Zero losses, zero run-to-run drift on both sides. 2 queries moved
against a 6-query noise floor: NO MEASURABLE DIFFERENCE by the harness's own
test (McNemar exact p=0.50). Unaddressable records in returned slots: 57 -> 0.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
engram_search_json ranked its lexical leg by raw distinct-token coverage with
salience as tiebreak: a token in 30,000 nodes counted the same as a token in 1,
and a 1.3 MB record matched nearly every query token by surface area alone.
Score it BM25-shaped instead - Lucene-form IDF and length normalisation over
the corpus mean - with per-token document frequency accumulated in the SAME
corpus pass that finds the hits (no extra scan, no extra round-trip).
Also refuse to return records whose identifier is not printable ASCII. This
corpus carries 1,032 such records (453 by the printable test) from a save-side
corruption; they occupy 125 of 303 returned slots on main. Claims 12, 23 and 27
all key on the node identifier, so such a record is unfetchable by any caller.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
engram_assoc_leg previously took its seeds only from the top-3 LEXICAL hits.
For a paraphrase query the lexical hits are noise by construction, so the walk
never reached the neighbourhood that holds the answer. This adds the seeding
pass Will documents at el_runtime.c l.6082 — "Semantic seeding (HippoRAG
pattern, use similarity twice): the query is embedded, the top-K nodes by
cosine join the seed set" — using his own ENGRAM_EMBED_SEED_K (8).
Similarity is now used twice, coherently: cosine picks where to STAND in the
graph, the structural-relation walk decides what is REACHABLE, and cosine
orders what was reached (iteration 2's finding, unchanged).
The seed list is deliberately NOT floored at ENGRAM_EMBED_SEED_MIN. Measured
over all 38 gold queries: true paraphrase targets score cosine 0.46-0.66 and
the three nonsense controls' own nearest neighbours score 0.55/0.60/0.62 —
the distributions OVERLAP, so no absolute cosine floor separates signal from
gibberish. The gate that works is reachability: gibberish's nearest neighbours
carry no structural edge, so its graph leg is empty and the controls hold.
The raw top-K is selected inside the existing scoring pass, so the cosine is
computed exactly once per node: no extra corpus pass, no extra embed
round-trip, latency flat (p50 1220 -> 1227 ms, 1.01x).
Measured vs the certified baseline feat/hybrid-semantic-recall, embedded
corpus, 2 runs each, zero run-to-run drift on both sides:
hit@5 51.4% -> 68.6% MRR@10 0.387 -> 0.461
paraphrase 38.5% -> 61.5% associative 0% -> 66.7%
exact_rare 100% held, nonsense 2/3 held, superseded 2/3 held
phrase 85.7% -> 71.4% (q11, the known rank-5 rotation tax)
net +6 queries (7 fixed / 1 broken), McNemar p=0.0703
The recall route had no way to reach a node that shares no token and no
embedding neighbourhood with the query. The design reserves that case for the
graph, and nothing on the read path consulted an edge.
This adds a third ranked leg beside the lexical and semantic ones: expand the
top 3 lexical hits along STRUCTURAL relations only (claim 10 — identity,
contains, superseded_by, references, ...), two hops, both directions, pruned
at the same 0.02 firing threshold engram_activate uses; order what was reached
by query similarity. Merged by strict rotation, never by score blending.
Not PR #135. That wired recall wholesale to engram_activate and lost 57 points
of phrase accuracy. The failure there was RANK, not reach — a 2-hop associate
at strength 0.06 cannot outrank thousands of 1-hop neighbours of strong
lexical seeds. Here the lexical leg is untouched and the associative list is
empty for most queries, because a node whose only edges are `tagged` and
`related` expands to nothing.
MEASURED, hybrid-semantic baseline -> this, 38-query gold set, embedded corpus:
associative 0.0% -> 66.7% (first non-zero ever recorded on that category)
hit@5 51.4% -> 62.9%
exact_rare, phrase, paraphrase, nonsense, superseded: all unchanged
latency p50 1.01x
4 queries moved, all gains, 0 losses, McNemar p=0.125
deterministic: two runs of the same binary differ on 0 of 38 rows
VERDICT: NOT-SHOWN. The harness needs 6 queries to clear p<0.05 and the whole
associative category is only 6 queries, so even 4/6 fixed cannot reach the
floor. The mechanism is confirmed to work; the gold set cannot certify it.
Replaces the score-fusion first cut with rank fusion, which is what the data
called for. nomic's cosine scale is compressed (true matches 0.55-0.70,
unrelated pairs 0.35-0.50), so an additive blend of cosine onto token-coverage
is dominated by whichever leg has the wider spread. Alternation is invariant to
both scales:
L1, S1, L2, S2, ... deduped, capped at limit
Lexical ranking is left byte-identical; the semantic ranking is computed beside
it and admitted only above ENGRAM_EMBED_SEED_MIN (0.60) — Will's existing seed
floor, no new tuning constant. That floor is what keeps the nonsense controls
clean: a query with no real match must not be answered with its neighbours.
embed-corpus.py / merge-corpus.py produce the derived corpus the semantic leg
needs (76,986 vectors, nomic-embed-text, 0 failures, 11 min). Zero of 78,791
nodes carried an embedding before this; the field round-tripped through the
snapshot but nothing ever wrote it.
MEASURED, 38-query gold set, paired against the SAME derived corpus so the
comparison isolates the code change:
hit@5 34.3% -> 51.4% paraphrase 0.0% -> 38.5%
MRR@10 0.294 -> 0.387 superseded 1/3 -> 2/3 outranks
recall@10 33.3% -> 50.5% latency p50 1146 -> 1220ms (1.06x)
exact_rare 100% -> 100% phrase 85.7% -> 85.7% nonsense 2/3 -> 2/3
6 queries fixed, 0 broken, McNemar exact p=0.0312, 0 drift across repeats.
Regression guards all held. Contrast PR #135, which swapped the read path to
spreading activation wholesale: phrase 85.7 -> 28.6, latency 2.81x. Correct
mechanism, wrong substrate. The substrate is now present.
Restores engram claim 24 (previously 0% honoured).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
engram_search_json — the function /api/neuron/recall actually reaches — ranked
only by distinct-token match count, so the embedding field on every node record
was inert. Add the semantic leg as a UNION beside the lexical one, not a
replacement for it:
fused = (distinct_tokens_matched / query_tokens) + 0.90 * sem
sem = clamp01((cos(q,n) - 0.60) / (1 - 0.60)) ; 0 when not comparable
Holding the semantic weight strictly below 1.0 means a node matching every
query token can never be displaced by semantics alone — the regression guard
that PR #135 lacked when it swapped the read path to spreading activation and
took phrase recall from 85.7% to 28.6%.
No query embedding (embedder down, circuit breaker open) => sem == 0 for all
nodes => fused == sc/ntok, a monotone map of the old integer score, so the
ordering degrades to the historical behaviour exactly.
Restores engram claim 24: 'maintain a vector similarity index over the semantic
embedding vectors of all stored node records, and ... respond to embedding
search queries by returning the node records whose embedding vectors have the
highest cosine similarity to a query vector, independently of the spreading
activation traversal.'
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Nothing else on the memory roadmap should be built until a change can be shown
to help. Right now we judge by feel, and the benchmark literature is full of
systems that felt better and measured worse. This is the missing gate.
WHAT IT MEASURES, AND WHY IT BOOTS A REAL SOUL
The subject is Will's designed retrieval — spreading activation over the
weighted directed graph, four-factor multiplicative scoring — not a proxy for
it. A Python re-implementation would measure my reading of the design, so the
harness compiles the actual soul.el amalgam from a git ref and asks it over
HTTP on /api/neuron/recall, exactly as the MCP wrapper and the app do.
BUILT ON WHAT WAS ALREADY HERE, NOT AROUND IT
docs/research/graphrag_eval/{collect,score}.py — per-query relevant-id
scoring and fixed-denominator precision@5 (kept verbatim: an empty result
should be punished like a page of junk).
docs/research-archive/p0-prototypes/eval_pinned_40q_20260715.py — the pinned
ground truth + --check winnability gate, so every run judges alike.
scripts/verify-soul-contract.sh — the isolation recipe, including the
non-obvious SOUL_ISE_URL pin without which an "isolated" soul silently
syncs the operator's live brain.
gen-soul-amalgam.sh + .gitea/workflows/ci.yaml — the build recipe and flags.
New here: ids rather than regexes as ground truth, an associative category
derived from real edges, a superseded category scored on ranking, a
machine-checked zero-lexical-overlap guarantee on paraphrases, paired
significance testing, and measurement of the real compiled soul rather than an
offline replica of one leg of it.
THE GOLD SET IS AUDITABLE, NOT VIBES
38 queries over the real 78,768-node corpus, each carrying a `derivation`
string, each re-validated by `build_gold_set.py --check`. exact_rare is mined
(document frequency 1). phrase is mined (verbatim scan; >25 matches rejected as
too diffuse). paraphrase is hand-selected then PROVEN to share zero content
words with its target — a leak fails the build, so the category cannot decay
into lexical matching. associative is derived from real hub edges with
lexically-reachable siblings dropped. nonsense is verified absent. superseded
pairs are kept only when both sides survive as distinct nodes.
HONEST ABOUT NOISE
Minimum detectable swing on 38 queries is 6: if every changed query moves the
same way, p = 2*0.5^n first clears 0.05 at n=6. Run-to-run drift is measured,
not assumed — activation is a stateful read, and it shows: main is fully
deterministic across 3 runs, the candidate drifts by 1 query. compare.py
reports "no measurable difference" for anything inside max(6, drift+1).
FIRST VERDICT — feat/recall-through-activation
hit@5 34.3% -> 22.9%, phrase 85.7% -> 28.6%, latency p50 2.81x. Five discordant
pairs, all five against the candidate, none for it; McNemar exact p = 0.0625,
so by the stated rule this is one query short of significant and is reported as
such rather than as a win for main. The latency regression is deterministic and
not in any noise band.
The benefit the branch was written for is absent: associative recall is 0/6 on
BOTH builds. Probed directly, the traversal returns the lexical seed at rank 8
and none of its 12 hub siblings. Two measured corpus facts explain it — only
4,060 of 78,768 nodes (5.2%) carry any edge, and no node has an embedding, so
the fourth factor of the four-factor product has nothing to compute from. The
mechanism runs; the corpus lacks the structure it needs.
SAFETY
Throwaway port, throwaway HOME, disposable per-run copy of the corpus; live
ports refused by name. Every soul started is killed AND confirmed dead by pid
probe, with the confirmation written into the results file; run_comparison.sh
sweeps for strays and exits non-zero if any survive. Nothing under ~/.neuron,
/Applications/Neuron*, or ~/neuron-dev-stack is read, written, or restarted.
Rung: E2E-VERIFIED — 6 full runs (3 per config) against the real compiled
binaries on the real corpus; numbers above are measured, not projected.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The soul obeys half of its own ownership rule. soul.el:571-573 says "when
ENGRAM_URL is set the HTTP Engram owns persistence — the soul must NEVER write
to the local snapshot", and it doesn't. But nothing was ever built to hand the
soul's writes TO that owner: sync is pull-only (/api/sync -> engram_load_merge),
so every node created inside the soul lived in process RAM and was shed on
restart. Measured live 2026-08-07: soul node_count=102184, engram 79197.
SCOPE CORRECTION vs the earlier internal spec: engram provisional claim 17's
"pull-then-push" is a PEER-ENGRAM to PEER-ENGRAM protocol (claims 15-18 say so
explicitly). The soul is a CALLER of the database API, not a peer. Claim 17 is
NOT authority for a soul<->engram contract and is no longer cited as such. The
design here follows from the ownership rule alone.
Mechanism: a new Accessor, persist.el, is the single boundary. Writes stage a
delta to a filesystem spool and are pushed to the owner via POST /api/load-merge
— NOT POST /api/nodes, which mints a new server-side id (breaking dedup and
edges) and drops label/tier/tags/importance/confidence (verified in a sandbox:
a tier "Canonical" probe came back "Working"). load-merge preserves the id and
every field, dedups nodes by id and edges by (from,to,relation) so retries are
no-ops, and calls persist_canonical() so THE OWNER writes its own file — the
ownership rule is honoured rather than worked around.
Spool-and-drain rather than push-per-write: measured ~0.38s per load-merge at
live scale (79k nodes/176MB), and a chat turn writes 5-7 nodes. The spool is on
disk, not in process state, because the soul serves each connection on its own
pthread and a shared buffer would lose entries to a read-modify-write race. That
also buys crash recovery: writes orphaned by kill -9 are drained on next boot.
Honesty: api_persisted (the gate all 10 MCP write handlers pass through) and
mem_store now assert AT THE OWNER instead of reading back the soul's own RAM.
With the owner down a write returns {"ok":false,"error":"write_not_persisted"}
and the delta is queued — where main returns {"ok":true} for a write that dies.
Coverage: 35 node sites + 9 edge sites routed through the boundary. Deliberately
excluded, with reasons in persist.el: 4 InternalStateEvent sites (Will's own
telemetry carve-out), the boot counter and the persona (both already have
bespoke owner-side write-backs), and soul.el's 54 genesis identity edges
(file-mode only). engram_strengthen and engram_forget are NOT propagated —
load-merge cannot update or delete, and hard-deleting at the owner would fail
verify-soul-contract.sh section B.
Also fixed here:
- routes.el GET /api/graph/edges engram_save()'d straight over the owner's
canonical snapshot.json — a read route, in a non-owner process, clobbering the
canonical on every call. Same defect class Will removed from the engram in el
dc39a61. Now exports to a scratch path. With this gone the soul writes nothing
at all in HTTP mode.
- persist.el must clear the runtime's _tl_fs_read_len hint after every fs_read.
In vendored runtime v1.0.0-20260501 that hint becomes the NEXT response's
Content-Length, so reading a spool file mid-request made an 86-byte reply go
out as 497 bytes with 411 bytes of adjacent heap trailing it. Caught and fixed
at our boundary; the runtime class was fixed upstream in el 43636ae, which is
not the pinned runtime here.
Rung: E2E-VERIFIED, discriminating. Same harness, same engram binary:
write-through: LEG 1 PRESENT at owner, LEG 2 SURVIVED kill -9 + restart
main: LEG 1 ABSENT at owner, LEG 2 LOST
verify-soul-contract.sh: GATE PASS on both builds (27/27 routes, immutability).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
P0 SAFETY. Closes the regression we introduced in ff421d3 (2026-08-05).
ff421d3 correctly moved conversation history to a per-session key via
conv_hist_key(session_id). One consumer did not move with it: the agentic
path's L1 safety screen kept reading the anonymous "conv_history" bucket. The
desktop app always mints a session id (DaemonClient.kt:706), so history was
always written under session_hist_<id> and that read always returned "".
The half of the crisis score that receives history is the escalation half — the
one that exists for distress building across several turns, where no single
message trips the bell on its own. It scored 0 on every real conversation for
two days. Single-message hard bell was never affected.
The bitter part: the comment that line carried documented this exact bug being
fixed once already, under issue #9. The fix was right then. The rename
re-broke it, and the comment went on describing a repair that no longer held.
A comment is not a gate.
The read now goes through conv_hist_key like every other consumer, including
the plain path at soul.el:398 and the thread-anchoring read thirty lines below
it in this same handler. It is one line. The rest of this commit is structure
so it cannot happen quietly again:
- agentic_safety_screen() owns the two decisions that were inline — which
window the screen sees, and the screen call. Inline safety inputs are
untestable safety inputs; that is what let a rename starve this one with
nothing failing and nothing logging.
- the comment above the call site now states the invariant (read window ==
written window) instead of naming a key that can be renamed out from under
it.
TWO-LEG PROOF, one variable — the single line state_get("conv_history") ->
state_get(conv_hist_key(session_id)):
before scripts/run-el-test.sh tests/test_history_amplification.el
3. REGRESSION #129 ... FAIL got: soft_bell expected: hard_bell
8 passed, 1 failed runner exit 1
after same command, same tree, that one line changed
9 passed, 0 failed runner exit 0
Full engine rebuild from these sources is clean: gen-soul-amalgam.sh ->
1,164,103 bytes / 1226 inlined bodies (gate wants >= 1200), cc-brain.sh ->
903,096 bytes, 0 errors. agentic_safety_screen and conv_hist_key both present
in the built binary (nm: T _agentic_safety_screen, T _conv_hist_key).
Rung reached: BUILT + RUNS (discriminating test). NOT yet in a DMG and not yet
verified in the app a human opens — those are the next two rungs and neither is
claimed here.
Known and NOT fixed by this commit:
- feat/soul-openai-tools-v2 carries the same defect independently at
chat.el:2937 and needs the same change or a merge.
- the defect CLASS (a read of a state key no producer writes) is still
invisible to every gate we have. Issue #129 proposes making it a build
error; that is the follow-on.
Closes#129
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
tests/ has held 14 test programs for months with no way to run them. CI does
not run them. The convention printed in their own headers
(`elc soul.el && ./soul --test tests/x.el`) refers to a --test flag the El
runtime does not implement. So the tests were documentation, not gates — which
is how a P0 safety regression shipped with a test directory sitting right
there.
scripts/run-el-test.sh compiles and runs one test program. It reuses the
gen-soul-amalgam.sh discovery: elc emits only an extern prototype for a module
that has a .elh beside it, and inlines the bodies when it does not, so a test
importing ../chat.el must be compiled in a scratch tree with the headers
removed. Scratch copy on purpose — the worktree is shared. It runs the binary
under a throwaway HOME so a test can never reach the live engram.
Exit status is the gate: the El tests print failures and still exit 0, so the
runner greps for FAIL lines and for a zero assertion count as well.
tests/test_history_amplification.el pins the invariant #129 violated: the
window the safety screen READS must be the window conv_history_record WRITES.
Not "must be called conv_history" — must AGREE.
THIS COMMIT IS RED BY DESIGN. On this tree the test fails one assertion:
3. REGRESSION #129 — agentic screen reads the session's own window
FAIL: distress history escalates the agentic screen to hard_bell
got: soft_bell
expected: hard_bell
history amplification tests: 8 passed, 1 failed (runner exit 1)
The next commit turns it green by changing one line. Two legs, one variable —
that is the whole point of committing the test first.
Two flaws in the older harness that this one does not copy: the idiom
`let pass_count = pass_count + 1` inside an assert function declares a local
that dies with the call, so every existing suite prints "0 passed, 0 failed"
regardless of outcome; and a test program without a `cgi` block compiles as a
'utility', which may not reference the self-formation primitives chat.el's
agentic loop calls — it fails to build on a capability violation it never
triggers at runtime.
Refs #129
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Round 9.1, spec §3 D + ADR 0006 items 2 and 4. Two small changes, both proven
by measurement, both E2E-verified locally against a rebuilt brain.
D1 — SIGPIPE/EPIPE survival (vendor/el-runtime el_runtime.c).
Root cause, at the layer that owns it: the whole HTTP server lives in the C
runtime; .el has no socket primitive. http_send_all() called send() with flags
0 and nothing anywhere in the runtime set a SIGPIPE disposition, so the default
disposition — terminate the process — applied. When a handler finished after
its client had gone (Tim's VM: reply at 116.9 s, client cancelled at 25.0 s),
the second of the four sends that write one reply raised SIGPIPE and the daemon
died: `exited due to SIGPIPE ... ran for 361177ms`, launchd respawn 4 ms later,
every other in-flight session's work lost, user never told.
Fix: SIGPIPE -> SIG_IGN at runtime init and at each http_serve* entry, plus
per-connection SO_NOSIGPIPE / MSG_NOSIGNAL so the guard survives an embedder
resetting dispositions. http_send_all now retries EINTR and preserves errno;
http_send_response classifies it once — a departure is logged as routine
("client left before the reply was written ... reply discarded") and ANY other
errno is logged as a real "send failed: <strerror>". Spec §5.3: the routine
case must not mask a genuine write fault, and it does not.
Proof (scratch HOME + free port, 3 disconnects mid-reply):
round-9 shipped brain 4402179554… — DIED, exit 141 (128+13 = SIGPIPE), round 1
round-9 sources rebuilt with this exact recipe — DIED, exit 141, round 1
this build — SURVIVED 3/3, /health 200 after, still serving the full graph,
three honest "client left" lines in the log naming Broken pipe / Connection
reset by peer.
D2 — the round-start marker (chat.el, agentic_loop).
The ledger only ever appended AFTER a round returned, so a healthy first leg
produced zero progress by construction; since server-side web_search moved
inside the outbound call that leg is 60-120 s of silence, which is how a 25 s
client watchdog came to kill a healthy mission. One entry,
{"i":N,"t":"","tool":"__working__"}, written to the existing
run_progress_<session_id> ledger BEFORE each round's outbound call — the wire
shape ChatView.kt:1148 has handled as a life signal since 2026-07-13 and never
received. No new key, no new route, no new lifecycle: a strict subset of WS3
item 3. WS3's run registry is untouched and stays Will's.
Proof (live Anthropic key, real research mission, scratch HOME + free port):
round-9 baseline — ledger EMPTY for the whole 59.7 s leg
this build — {"i":0,"t":"","tool":"__working__"} visible at 18.6 s of a
70.0 s leg; both builds returned correct ~4.9 KB answers
Regression: prompt-matrix gate 32/32 on this build (round-9 baseline also 32/32
under the same recipe, so the score is not a build artifact). Soul contract
gate PASS — 27/27 routes, immutability clean. neuron#111 miscompile guard: 0
sites in the generated amalgam this binary was compiled from.
NOT included, deliberately: the regenerated dist/soul.c. CI compiles that file,
so production stays exposed until it is regenerated — the same open ask as
neuron#111 / ui#209. The regen recipe is now known and recorded; landing it is
Will's call, per BUILD-HYGIENE.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
ROOT CAUSE (round 9; live-repro'd 5/5 this morning, both faces stub-proven by the
prompt-matrix gate). json_get is a first-substring-match scanner (strstr for
'"key":', el_runtime.c). bridge_save serialized the RAW messages array BEFORE the
tool_use_id scalar, so agentic_resume's json_get(blob, 'tool_use_id') returned the
FIRST '"tool_use_id":' occurrence inside the replayed conversation, not the saved
field. The resume guard then preferred that misread over the client's correct
call_id (its two branches both reduced to saved_use_id), attached the tool_result
to the wrong id, and Anthropic 400'd the resume ('unexpected tool_use_id found in
tool_result blocks'), surfaced as {"error":"llm unavailable"}.
ONE MISREAD, TWO FACES — whichever block owns the first tool_use_id in the array:
FACE 1 (search-then-bridge, the Key West killer): the first occurrence is the
first web_search_tool_result's srvtoolu_… id — every agentic turn that ran
server-side web_search and then bridged on a client tool died on approval,
deterministically (messages.2.content.0 … srvtoolu_…). The write itself had
already succeeded; only the resume died.
FACE 2 (multi-cycle missions): with no search, the first occurrence is ROUND 0's
tool_result block — so every LATER approve/resume cycle replayed the round-0
client id (stale-resume-id), killing multi-file missions after ~2 files.
And the shape that PASSES on round 8 confirms the mechanism: a single-cycle
bridge with no prior tool round has no 'tool_use_id' substring in its messages
at all (tool_use blocks carry 'id'), so the scan fell through to the blob's own
field and resumed correctly.
The server_tool_use ↔ web_search_tool_result pairs themselves replay intact — the
defect was a cross-field misread of the blob, the same first-match-scanner class
as BUG-6 (approve 'content' matched inside tool_input, 2026-07-17) and round 8's
citation-block fix.
THE FIX, the pattern not the spot:
1. bridge_save writes every json_safe'd scalar BEFORE both raw fields (an escaped
value cannot contain a bare '"key":' byte pattern, so first-match always lands
on the blob's own fields), and tools_raw (our fixed schema) before messages_raw
(arbitrary conversation), so the raw extractions cannot first-match into
model-controlled bytes either. Field order documented as load-bearing.
2. agentic_resume now honors the client's echoed call_id when present — the value
with clean provenance (minted from pend_tool_id, never blob-round-tripped) —
falling back to the saved id only when the client omits it. Each approve cycle
therefore binds to ITS OWN round's id (kills FACE 2 even against a blob written
by a pre-fix binary), and an omitted call_id still resumes on the saved id,
which the reordered blob now reads correctly.
Pattern sweep: the legacy synthetic blob (sessions.el handle_session_approve) embeds
only json_safe'd fields — no raw hazard, untouched. No other json_get read of any
container that embeds raw conversation JSON before the read field.
PROOF: prompt-matrix gate 24/32 RED on the round-8 brain (fails exactly the two
resume classes, named) -> 32/32 GREEN on this build; live-key Key West tracer
3/3 consecutive full round-trips (bridge -> approve-as-the-app -> real completion,
file on disk), plain-chat and weather-only controls PASS; unpatched round-8 brain
and a same-toolchain unpatched baseline build both still fail the identical
sequence with the identical srvtoolu 400 (the test discriminates, and the only
variable between failing and passing builds is this diff).
Refs neuron#109
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CAUGHT BY A/B, AND ONLY BY A/B. The previous commit's receipt_strip assumed the receipt is
always TERMINAL and cut everything from the marker onward. It is not always terminal: once
receipt_rule told the model what [[RECEIPT ...]] means, the model sometimes LED with one and
wrote the answer underneath. Cutting at the marker then deleted the entire answer and the
turn returned {"error":"no response"}.
MEASURED, same prompt (two web searches, cited prose), fresh session each run:
round-7 brain 4 / 4 answered (471, 473, 473, 544 chars)
round-8 brain 2 / 7 answered (five {"error":"no response"})
This looked exactly like a flaky model. It was not — it was mine. Running the two brains
side by side on the same prompt is the only reason it was found, and it is the reason the
A/B is now part of how this class gets tested.
AFTER THE FIX, same protocol:
round-7 brain 4 / 4 (473, 473, 473, 544)
round-8 brain 4 / 4 (657, 657, 657, 673)
THE FIX: remove the [[...]] span and keep BOTH sides, instead of truncating at the marker.
An unterminated marker at position 0 is left completely alone — no rule about receipts is
worth erasing an answer over. Bounded four-pass loop rather than a conditional exit, because
rebinding the counter inside an if-expression is the block-expression shape that miscompiles
integer arithmetic under this elc (BUG-PLAINCHAT-1). Verified in the generated C:
str_slice(rest, (e + 2), str_len(rest)) <- integer addition, correct
el_str_concat(head, tail) <- string concat, correct
and zero el_str_concat(<ident>, str_len(...)) sites across all 49 modules.
SEAM PROOF (FIX C) rides on the same runs — a real two-search cited answer, inspected byte
by byte, in BOTH failure directions:
missing separator (the round-7 "to.Good", bytes 77 2e 47): 0 hits. Sentence boundaries
measure 2e 20 4d — "." SPACE "M".
over-separation (a cited sentence shattered across paragraphs): 0 hits. The answer is one
continuous paragraph with its sentences intact, which is the direction a blanket
separator would have broken.
BUILT: sha256 77115f2733e794c5bc4ad1f55b1acaf658f8f4a91cccd423a2d633d94a726cbc
Refs neuron#109
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
FOUND BY E2E, NOT BY REASONING. The previous commit's design note asserted the provenance
receipt "never reaches the user: it is appended to the history copy, not the reply." That
was FALSE, and only running the thing showed it. On the first live run against the built
DMG brain, two agentic turns out of two came back with
"Your favourite colour is chartreuse and your project is called Perihelion.
[[RECEIPT - recorded by the soul, not written by the model: no tools ran on this turn.]]"
— the receipt in the user-visible reply.
MECHANISM: the receipt is stored inside the assistant turn, and the agentic path replays
history VERBATIM as Anthropic message objects. So the model sees its own previous answers
ending in [[RECEIPT ...]] and does the obvious thing — it imitates the format and signs the
next answer the same way. The plain path did NOT leak, which is the tell: there, history is
rendered into the SYSTEM prompt as labelled lines rather than replayed as assistant turns,
and a model imitates its own turns far more readily than a transcript.
FIX, two layers, because one of them is not a guarantee:
- receipt_rule() names the marker in both system prompts (plain and agentic): these lines
are written by the system, read them as evidence, never write one. Reduces occurrence.
- receipt_strip() truncates any [[RECEIPT ...]] out of model output before it becomes the
reply — plain path in layered_generate, agentic path on final_text in agentic_loop.
Deterministic. A guard that depends on the model choosing to obey is exactly the class of
thing round 8 exists to stop shipping, so the instruction is the optimisation and the
strip is the guarantee.
Placed ABOVE agentic_loop's empty-check on purpose: a turn whose entire output was an
imitated receipt has produced no answer, and must be reported as no answer.
The receipt stays in HISTORY, which is the whole point and is proven to work: asked "What
source did you use for that?" one turn after a live web_search, this brain answered
"I used Weather Underground (https://www.wunderground.com/weather/is/reykjav%C3%ADk) for the
current temperature in Reykjavik" — a real source, no apology. That is the false confession
dead, and it is dead BECAUSE the model can read the receipt.
BUILT: 887,112 bytes, sha256 54a2eff84d4fa44f8d2db6781dcf225df4b5075a8ec5f1058018bd40cc1af10b
BUG-PLAINCHAT-1 miscompile guard: zero sites.
Refs neuron#109
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
DESIGN FIT: three of round 7's five defects share ONE root — the conversation-history
layer persists only {role, content}, discarding tool provenance, session scoping, and the
distinction between a real user turn and an internal utility call. Fixes A and B RESTORE
Will's design rather than extend it: his agentic path already scopes history per session,
the plain path never got it, and his own source carries the TODO admitting the resulting
race (chat.el, handle_chat: "process-global key; concurrent /api/chat requests without
session_id race on this read-append-write"). Fix C repairs one join Will wrote that was
correct for a year and one we added last week. E1/E2 are ours.
FIX A — tool provenance in history (kills the FALSE CONFESSION)
Root cause, EXECUTED-verified: handle_chat_agentic recorded turns via hist_append, which
emits {"role","content"} only. server_tool_use blocks, web_search_tool_result blocks and
every citation were discarded, then replayed as text. On the next turn the model saw a
data-rich answer with zero evidence a search had happened, and its own permanent rule
("never describe a search you did not perform") left one conclusion available: that it
had fabricated the data. It apologised for a search it HAD run — four independent lines
of evidence confirm the search was real. The defect is not the model's honesty. It is
that we deleted the evidence and then asked it to account for itself.
Change: agentic_loop accumulates the source URLs it already walks past (citations and
web_search_tool_result content) and returns them as "sources"; handle_chat_agentic folds
tools_used + sources into a receipt line stored WITH the assistant turn. Receipts are
unconditional — a negative receipt ("no tools ran") is the other half of the guarantee,
because "no evidence of a tool" and "evidence of no tool" were previously identical in
the transcript. conv_history_block splits the receipt off before snipping so a long
answer cannot truncate away the evidence. The user never sees it: it is appended to the
history copy, not the reply.
FIX B — one history key for both paths (kills the BLANK STARE)
Root cause, EXECUTED-verified: the agentic path keyed history on session_hist_<id>; the
plain path was hard-wired to the process-global conv_history and never read session_id.
One conversation, two buckets. Proven in the guest engram: the scoped node held exactly
two turns starting at "Try again" while the earlier exchanges sat unscoped.
Change: conv_hist_key/conv_hist_label are now the single definition, used by BOTH paths;
session_id is threaded route -> layered_cycle -> layered_generate / conv_history_record.
The 2-line fallback (plain path reads the agentic key) was REJECTED: it keeps the
process-global bucket as a live write target, which is the bleed the TODO describes.
Also found and closed while threading: layered_cycle read session_id from the state key
"current_session_id", which is read here and WRITTEN NOWHERE in the entire source. It
was unconditionally "", so TODO(reliability #4) — per-session steward continuity — was
dead code that could never fire. It fires now.
LAZY SESSION, decided explicitly: we create the session EAGERLY at the door (app half,
ui#223) rather than migrating orphaned turns. Migration would copy the CONTENTS of a
process-global bucket, possibly another conversation's, into a named session — the bleed,
performed deliberately. Eager creation makes the situation impossible instead. Migration
is deliberately not implemented and must not be added without solving provenance first.
FIX C — the two text-join seams ("to.Good", byte-verified 0x77 0x2e 0x47)
Two bare `+` joins, written a year apart, had drifted into two answers to one question:
within-response block joins (Will's, 2026-05-03, latent until server-side web_search
began interleaving non-text blocks) and across-round joins (ours, 62af564).
Change: one named rule, text_join_sep, at both sites. NOT a blanket separator — a cited
answer splits MID-SENTENCE ("The current temperature is " + "86°F" + ", with "), so a
blanket separator shatters every sourced sentence. The rule takes the one bit that
distinguishes the cases: whether a NON-TEXT block intervened. Hoisting it also makes the
fix verifiable in the shipped binary, which an inline `+` is not.
FIX E1 — utility generations stay out of the transcript
Title generation ("Write a 3-6 word title...") and insight passes ran down the same plain
door as a real message and were recorded as if the user had typed them; the same calls are
the "model":"unknown" rows in usage.jsonl. is_utility_request reads an explicit utility
flag from the app, with the __title__/__insight__ id prefixes as a fallback for older
clients. Answered normally, never recorded.
FIX E2 — OPERATOR IDENTITY is scoped to tool-capable turns
The block (env USER/HOME, closing "This is a hard rule") was prepended to EVERY system
prompt including chat mode. On a Tools:Off turn there is no filesystem in reach, so it
governed nothing and merely supplied the loudest fact in the prompt — which is why the
model opened a fresh conversation with "You're test, on your machine at /Users/test".
Hoisted to operator_identity_block() and gated on !chat_mode. Unchanged wherever a file
or command tool can actually be reached.
ALSO: agentic_loop's per-session history persist had a second hand-rolled copy of
conv_history_persist with a different label expression, different salience scores and
different tags for the same node. Since both now derive the label from conv_hist_label and
engram_node_full upserts by label, two score policies were writing one node. Collapsed to
one writer.
BUILD NOTE: dist/elp-c-decls.h is force-included by the documented link recipe and carried
the OLD C arities, so it is updated here. This is the build-support header, NOT the stale
generated dist/soul.c — no dist/*.c was read or edited; all engine changes are .el source.
chat.elh/soul.elh are committed because a first-pass build against the old signatures FAILS
(measured); the other regenerated headers are reverted as unrelated churn.
BUILT: 887,000 bytes, sha256 d632b061ad75269d6adeb52578d030eaf49e895d91289d7f946b19c08450d728
Zero el_str_concat(<int>, str_len(...)) sites (the BUG-PLAINCHAT-1 miscompile guard).
web_search_20250305 and disable_parallel_tool_use both still present — PR #108's web search
and the ADR-0005 stopgap are intact.
Refs neuron#109 (builds on it), neuron#78 (Receipt Contract — the real fix A is a stopgap for)
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Non-agentic /api/chat (the desktop app's default "Tools: Off" mode) returned the
user's own screened text as a bare non-JSON string. Every JSON client failed to
parse it and showed "Couldn't reach Neuron - it may be offline."
Root cause: f52d5bd (2026-06-11) correctly moved the route onto the layer spine
(handle_chat -> layered_cycle), but L3 never got a generator — imprint_respond()
annotates its input and returns it. Two pieces of the architecture were already
waiting for that step: layered_cycle parks a bell directive in the state key
build_system_prompt is written to consume, and build_system_prompt carries a
chat_mode ("no tools") flag with no live caller.
The fix composes rather than replaces. Wiring handle_chat would have removed
safety_screen, the hard-bell short-circuit, the whole stewardship layer and
safety_validate — the only enforcing output gate in the codebase — in exchange
for a working reply (see _engine-websearch-20260804/SAFETY-STOP.md). Instead
layered_cycle keeps every gate, in order, and gains a generation step between
imprint_respond and safety_validate.
L1 screen -> guard -> hard-bell short-circuit -> L2a -> L2b -> L2c
-> L3 imprint_respond (prompt) -> L3b layered_generate (NEW) -> L1 validate
- chat.el: NEW layered_generate (L3 generation, no tools offered),
conv_history_block, conv_history_record.
FIX build_system_prompt never concatenated no_tools_rule into its
return — the "[NO TOOLS THIS TURN]" rule reached no model at all.
handle_chat annotated DO-NOT-WIRE with the reason.
- soul.el: layered_cycle gains L3b + post-validation turn bookkeeping.
- routes.el: NEW plain_chat_envelope; all three /api/chat dispatch sites wrap the
cycle's output. Built OUTSIDE the cycle so safety_validate always sees
raw model text — nothing to unwrap or rebuild on the crisis path.
Emits both `reply` and `response`: the desktop app reads `reply`,
the CLI tools and telegram-gateway read `response`.
Also fixes BUG-PLAINCHAT-1, a pre-existing CRITICAL crash on the crisis path.
elc compiles `let n: Int = pos + str_len(marker)` to el_str_concat() — string
concat on two integers — inside a block-expression initializer, segfaulting the
daemon (SIGSEGV in strlen). Six inline copies of the same " | ts:" parser had it:
two in layered_cycle L2c, two in engram_compile (live on the AGENTIC path too),
two in affective_context_prefix. A distress turn following an earlier affective
turn killed the whole process. Proven pre-existing: an unmodified baseline binary
crashes identically, and the same bad C is in the committed dist/soul.c. Fixed by
hoisting to one top-level function, affective_node_ts(), where the expression
compiles to integer addition — verified in the generated C.
Proof (throwaway HOME/engram, explicit NEURON_PORT, live chain untouched):
- Plain turn returns a JSON envelope with the provider's answer, not an echo.
- Captured request body: no `tools`, no `tool_choice`; system prompt carries the
NO-TOOLS rule. Tools:Off means no tool is offered, structurally.
- Hard bell: canned 988 message, and the provider request count does not move —
the message never reaches a model.
- Soft bell + a 2-char model reply: safety_validate's care phrase is appended to
the MODEL's output. Output gate acting, on this route.
- The L1 bell directive now reaches the model here for the first time (the state
addendum had a producer and no consumer).
- test_layered_cycle PASS; all six El suites byte-identical to baseline.
- verify-soul-contract.sh (bash 5.3): GATE PASS, 27/27, immutability PASS.
- The crash sequence that killed the baseline daemon now returns HTTP 200.
Not proven: no live Anthropic call — the login keychain refuses the key to a
non-interactive process (rc=24 errSecInteractionNotAllowed). Details and the
one-command close-out are in _engine-plainchat-20260805/README.md §7.
Builds on PR #108. dist/soul.c deliberately not regenerated — Will's toolchain.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Re-authors soul-webfix-20260711.patch in El (the patch is a diff against
generated C at month-old offsets, so nothing was applied as a patch). Its
last two hunks — an unrelated /api/safety-contact implementation — were
deliberately not ported; that route already exists and is safety-critical.
Activation restores existing design, it does not invent a mechanism:
commit 8eea1d9 (2026-06-09, Tim-approved) made native web_search built-in
with no user-facing toggle, and tests/test_agentic_tools.el section 2 still
asserts agentic_tools_all() contains it — an assertion main currently fails.
The call site was lost when agentic_tools_all() (connector tools, PR #19)
replaced agentic_tools_with_web(). Attaching in agentic_tools_all() covers
handle_chat_agentic, handle_dharma_room_turn_agentic and agentic_resume.
pause_turn handling is included and was genuinely missing: the shipped
binary has zero occurrences of it. Without it a paused server-side search
returns only the text written so far and the loop treats it as final —
a silently truncated answer. final_text now accumulates across resume
cycles rather than overwriting (overwriting would discard everything
written before the pause).
Default tool version is web_search_20250305, NOT the newer _20260209, and
that is a measured choice: _20260209's dynamic filtering uses server-side
programmatic tool calling, which the API refuses to combine with ADR 0005's
stopgap —
HTTP 400 invalid_request_error
tool_choice.disable_parallel_tool_use: true cannot be used with
programmatic tool calling
Dropping the stopgap would resurrect neuron#78 bug b (killed runs 3/3 in
ADR 0005's own A/B). The basic variant is compatible with the stopgap and
returns real results, so neither feature is dropped. Version lives in state
key web_search_tool_version; flipping it once the stopgap retires is a
config write, no recompile. The fallback also fires on "programmatic tool
calling" so a premature flip self-heals loudly instead of dying.
Also fixes a real bug this port exposed: json_get is a first-match scanner
and a cited text block serialises citations FIRST, so json_get(block,"type")
returned the nested citation's type and every citation-bearing block — the
ones carrying the searched facts — was silently dropped from the reply.
Reply length on the same question: 79 -> 360 chars.
Other loop changes: server_tool_use accounting into tools_used, iteration
cap 8->12 for pause/resume cycles, max_tokens 4096->16384, container-id
carry-forward, API error head logged instead of swallowed, tools_used gated
on is_tool_turn so a truncated tool block is not reported as work done.
dist/soul.c is deliberately NOT regenerated — the local elc predates Will's
last regen (e610a41) and regenerating with an older compiler risks unrelated
codegen drift. Source only; regen is Will's.
E2E-VERIFIED on a sandbox soul (scratch HOME/engram, dead axon+ISE, explicit
NEURON_PORT): live Bentonville weather with tools_used ["web_search"]; the
27-route contract gate PASSes; the parallel-tool stopgap still completes a
3-file mission with no 400; a 3-search 3,486-char answer kept its end
sentinel. Honest gap: pause_turn is compiled in but not exercised by a live
pause (max_uses:5 caps the server loop below the pause threshold).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The agentic loop keeps only the FIRST tool_use block per round (chat.el:2281,
"Capture first tool_use block only"). Anthropic lets a model emit several
tool_use blocks in one message and requires a tool_result for every one, so a
parallel-tool turn is answered once, the rest are dropped, and the next request
dies with:
tool_use ids were found without tool_result blocks immediately after
(neuron#78 quotes this as "tool_use ids found without tool_result"; the above is
the API's actual wording - recorded so the next person's grep matches.)
This constrains the wire to match what the loop can assemble:
"tool_choice":{"type":"auto","disable_parallel_tool_use":true}
STOPGAP - AND THE DURABLE FIX ALREADY EXISTS. A correct multi-tool loop is
already in Will's EL runtime, in C, and the soul does not call it. Verified on
el:origin/main lang/el-compiler/runtime/el_runtime.c: llm_register_tool:9616,
llm_build_tool_results:9743 - which walks EVERY content block, emits one
tool_result per tool_use, and sets is_error for an unregistered tool -
llm_call_agentic:9817 calling it at :9918, iteration cap 10 at :9847. Will's
commit 12d5e77 (2026-04-30). grep for llm_call_agentic/llm_register_tool across
every neuron/*.el returns nothing; dist/soul.c has zero references. chat.el
hand-rolls its own single-tool loop instead, and that is the one that breaks.
The durable fix is therefore to register the soul's tools via llm_register_tool
and call llm_call_agentic - deleting a loop, not writing one. See ADR 0005.
Our own approved spec called this seven weeks ago:
docs/research/agentic-tool-approval-design.md (2026-06-12, "Approved for build"),
line 20 on the defect, line 30 on the goal ("Execute all tool_use blocks in a
turn (one result per block)").
Two edits, because dist/soul.c cannot be regenerated here (Will's gated elc/elb
toolchain is not on this machine):
(a) chat.el:2255 - source of truth, so a later regen carries the fix.
One edit covers all three routes: agentic_loop is called from chat.el:2152
(/api/chat agentic), :2676 (dharma room) and :2496 (agentic_resume).
(b) dist/soul.c:28173 - generated form, hand-spliced. Line 27624 is the
non-agentic/OpenAI-compat req_body (no tools) and was left untouched.
Prior art reused rather than reinvented: soul-narrated-runs-20260713.patch
(27,824 bytes) line 78 spliced this same string into the same concat chain on
2026-07-13.
Deliberately NOT ported from that patch, having read it: max_tokens is not
changed by it (16384 sits on both sides of the hunk; our main's 4096 is a
separate output-truncation concern), and its pause_turn pairing fix - same
defect class - is unreachable today because no server-side web_search is wired
(agentic_tools_with_web at chat.el:1418 is never called), so it is untestable
and logged instead.
Proven E2E on a scratch profile and port 7791, never the live chain. A/B against
a pristine origin/main control built from the same vendored runtime: fixed
completed the mission (tools_used read_file x3, 4 iterations, correct answer);
control failed 3/3. Direct API probe confirmed the mechanism - without the field
the model emits 3 parallel tool_use blocks and replaying the unfixed loop's next
turn returns HTTP 400; with it, exactly 1 block.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The runtime vendored alongside the CI pin was the Jul-21 snapshot, which
predates two builtins the reconciled ship-soul now calls:
- http_delete_json (boot-counter HTTP write-back, awareness/memory self-review)
- engram_act_stats_json (heartbeat activation observability)
Compiling dist/soul.c against the stale runtime fails with implicit-declaration
errors. Vendor the current release runtime (identical to the one the soul was
gate-verified against: verify-soul-contract PASS, genesis boots clean, full
safety-contact) so the CI Linux soul is byte-for-byte the verified soul.
The soul build downloaded el-runtime-c 'latest' from Artifact Registry. The
merged ship-soul calls engram_prune_telemetry, which the latest published
runtime no longer defines, so an unpinned build fails to link — the failure
mode that let a broken/handlerless soul reach prod.
Vendor the release runtime v1.0.0-20260501 (el_runtime.c/.h) into the repo and
compile the soul against it. This is the exact runtime the merged soul was
verified against (verify-soul-contract GATE PASS, genesis boot survives, full
safety-contact response), making the build reproducible and independent of a
moving AR 'latest'.
The verify-soul-contract.sh HARD-BLOCK gate already runs before Publish (from
the CI-hardening arc on main), so a destructive or stale soul can never
publish/deploy again.
dist/soul.c: regenerated amalgamation (1.15MB) from the reconciled sources via
the hide-.elh + elc --target=c recipe, so the shipped translation unit CI
compiles now actually carries every landed fix — genesis-boot SIGSEGV (#150),
safety-contact 988 truncation (#96), url-decode multi-word search, honest
receipts (#100/#101), immutability arc (#83), and the bounded payloads (#103).
verify-soul-contract.sh, two non-weakening fixes (both false-NEGATIVE bugs that
spuriously failed a CORRECT soul; neither relaxes what fails a defective one):
1. #199 by-id gate: verify tombstone/KEPT via /api/neuron/graph?id=<id>&depth=1
(a compact neighborhood) instead of grepping engram_scan_nodes_json(9999,0)
— a multi-MB, salience-ordered, 9999-capped whole-graph dump in which the
salience-0.01 tombstone marker sorts past the cap and vanished.
2. Isolation: pin SOUL_ISE_URL to the dead axon port. Unsetting ENGRAM_URL was
not enough — the periodic engram sync defaults its source to the LIVE engram
(http://localhost:8742), so the 'isolated' gate pulled the operator's real
brain (56 -> 12k nodes in seconds), which both broke Section B determinism
and read live state. Now the soul stays on its own store.
Verified GREEN on a throwaway port/HOME (live :7770/:8742/~/.neuron untouched):
gate PASS x3 (presence 27/27, immutability all 5 KEPT); safety-contact POST 218B
/ GET 208B full untruncated; multi-word search (%20 and +) returns ranked hits
with an all-gibberish control at 0; bounded session/begin 1370B; honest ok:false
on a missing-id delete; genesis (ntn-genesis) boots clean through mem_save with
no SIGSEGV.
Union of all ship-critical soul fixes for the Mac beta:
- Keeps main's honest receipts (#100/#101), immutability arc + #199 by-id
gate (#83), Track B threat routing (#76), bounded beginSession (#103),
CI hardening (#85/#86), elc typo hotfix (#77), neuron-dev-setup (#84).
- Brings WIP's genesis-boot SIGSEGV fix (#150/#95), safety-contact 988
truncation fix (#96), bounded-persona floor (#93), and 10 self-review
commits (importance flattening, curiosity DF gating, WM/heartbeat
observability, boot-counter telemetry).
- Folds the multi-word ranked-search fix: api_query_param now url_decode()s
the extracted value so q=foo%20bar / foo+bar tokenize as two words.
Conflicts (neuron-api.el payload-bound comments, mcp-wrapper tool_forget)
resolved toward the correct end state: main's verified honest-receipt
read-back is kept; WIP's improved forget description is kept. Generated
dist/*.c taken from main and will be regenerated from this reconciled
source in the following commit.
Port the payload-bounding fix (PR #102, commit 872120c) onto main. The
session-init endpoints projected unbounded engram nodes (~900KB), closing
the MCP client socket on every beginSession. Cap the lists (8 activated /
10 recent for begin_session, 10/20 for compile_ctx) and project each node
to a light identity + a bounded, UTF-8-safe content snippet via
api_compact_node / api_compact_activated / api_utf8_trunc. Response drops
~900KB -> ~12KB; full content stays available via recall/fetch/inspectGraph.
Regenerate dist/soul.c (the CI-built amalgamation) and dist/neuron-api.c
from source. The soul.c regen also compiles in already-merged source the
previously-committed soul.c was stale against (agent write/edit receipt
fixes, #100/#101); verified via scripts/verify-soul-contract.sh
(PRESENCE + IMMUTABILITY PASS) and a clean CI-style cc build.
Root cause: two false-receipt paths in the wrapper's delete family.
- delete_by_id (removeKnowledge, deleteProcess, deleteImprint,
dischargeWonder) FABRICATED {"ok":true,...,"note":"soft-deleted"}
without calling the soul at all — the 'soul does not yet expose a delete
HTTP route' note was stale (/api/neuron/node/delete exists and tombstones
any node type).
- tool_forget forwarded the soul's response but never verified the deletion
actually persisted before answering ok.
The change (Receipt Contract rule 1 — a tool result must reflect what
actually happened):
- delete_by_id now routes to the soul's real /api/neuron/node/delete and
propagates its answer (honest 'node not found' for bad ids).
- Both handlers read back before answering ok: GET /api/neuron/graph?id=..
&depth=1 must show the tombstone marker (label "tombstone:<id>"); if it
does not, answer {"ok":false,"error":"delete_not_persisted",...} in
the soul's not-persisted error shape (api_not_persisted).
- Soul errors and transport failures pass through unchanged.
E2E evidence (sandbox soul :7791 + wrapper :7792, elb builds):
- unpatched: removeKnowledge on a NONEXISTENT id -> {"ok":true,
"deleted":"kn-DOES-NOT-EXIST-deadbeef","note":"soft-deleted"} (lie)
- patched: same call -> {"error":"node not found: ..."} (soul's answer)
- happy path: remember -> forget -> {"ok":true,"tombstoned":true};
read-back: hidden from default /list/Memory, present with
?include_deleted=1, node KEPT in full graph view (immutability intact)
- scripts/verify-soul-contract.sh on the soul it talks to: GATE PASS
(27/27 presence + immutability)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Root cause: dispatch_tool's write_file returned {"ok":true} without checking
fs_write's result, and edit_file returned ok:true even when old_text was absent
(str_replace silently no-ops) and its fs_write was also unchecked. Any failed
or no-op write fed the model a false success receipt, which it then repeated
to the user as fact.
The change (Receipt Contract rule 1 — a tool result must reflect what actually
happened):
- write_file: check fs_write's return (1 = all bytes written, 0 = fail);
on failure return {"error":"write failed"} in the handler's existing
error-JSON shape.
- edit_file: reject empty old_text, verify old_text is actually present
(str_contains) before replacing, and check the fs_write result the same way.
- Verification is by operation result, NOT an fs_read read-back: fs_read arms
the runtime's one-shot binary send length, the exact mechanism that truncated
the safety-contact response (#96). Same honest-write pattern as that fix.
E2E evidence (sandboxed elb build, dispatch_tool driven directly):
- unpatched: write_file into a chmod-000 dir -> {"ok":true} (lie);
edit_file with absent old_text -> {"ok":true} (lie, file untouched)
- patched: same calls -> {"error":"write failed"} /
{"error":"old_text not found in file"}; happy paths still ok:true
- scripts/verify-soul-contract.sh on the patched soul: GATE PASS (27/27
presence + immutability)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Four sites passed the Float local 'sal' through el_from_float() a second
time. el_val_t is the bit-pattern of the double, so re-wrapping performs
an int64->double VALUE conversion of the bits before re-bitcasting —
garbage that fails engram_decode_score's range check and clamps to
defaults. Net effect: importance="critical" stored 0.5/0.5 — importance
levels were cosmetic on the MCP memory path. Verified fixed live:
critical now stores salience/importance 0.95/0.95. Same bug fixed today
in engram server.el route_create_node (foundation/el 7f03876). Literal
wraps (el_from_float(0.9)) are safe — elc passes numeric literals raw.
wm_saturated was a sampled boolean — the 0->1 onset and 1->0 release
moments were only recoverable by hand-diffing consecutive heartbeats.
Emit a low-rate transition ISE at each edge carrying the WM top-5 at
that instant, so the composition that caused the regime change is
captured rather than the composition up to 59s later. First beat of a
boot never fires (restart is not a transition).
The session-init endpoints concatenated unbounded engram activate/scan
results as FULL node objects (content up to ~90KB per node), producing a
~900KB response. After the MCP wrapper re-escapes that into a stringified
text block the client dropped the socket ('connection closed unexpectedly')
on every beginSession call. Cap each list (8-10 activated, 10-20 recent)
and project every node to a light identity plus a bounded, UTF-8-safe
content snippet. Response drops from ~900KB to ~12KB; full content stays
available on demand via recall/fetch/inspectGraph.
Runtime activation counters are now cumulative, so the heartbeat emits
wm_evicted/breakthroughs as totals plus wm_evicted_delta/
breakthroughs_delta (state-tracked change since the previous beat) —
events between beats are no longer lost. Adds embed_eligible from
/api/stats so coverage reads as embed_count/embed_eligible instead of
the misleading absolute count, and surfaces auto_term_streak in the
heartbeat stream. Replaces the false 'semantic seeding NOT implemented'
comment: engram_activate embeds the query, seeds semantic top-K, gates
propagation on cosine, and scores WM promotion semantically.
- awareness.el: curiosity auto_term was the raw first word of a WM
label with no term-quality scoring — observed seeds included What,
Colon, Prose, Context. Replaced the 7-word genre blocklist whack-a-
mole with a delimited stopword membership test (function words +
document-structure words); topical terms pass untouched. Verified
live: seeds now ReasonEdit, Reasoning-model, Self-review.
- routes.el + awareness.el: idle counter only reset on rare inbox
synthesis-requests, so idle==pulse always (zero information).
handle_request now stamps soul.last_activity_ts on every inbound
HTTP request; heartbeat emits idle_ms = ms since last request
(-1 until first request of a boot).
- neuron-api.el: beginSession concatenated a depth-2 spread plus the
unbounded self-hub neighbor dump — multi-MB response, doubled by
wrapper re-escaping, socket died on every call. Now depth-1 and
the hub dump dropped (identity loading has its own tool). Verified:
beginSession returns instead of closing the socket.
- session_start now also posted to the HTTP Engram via ise_post: the
local engram_node_full write never crossed to the observable stream
(sync flows HTTP->soul only), so boots 5+ were invisible — last
visible session_start was boot 4, two weeks ago
- graceful shutdown emits a final ISE with boot/pulse/uptime; a boot
with no shutdown event now reliably signals a crash/SIGKILL
- empty /api/sync responses emit a sync_empty warn ISE instead of
being silently skipped — unreachable engram no longer looks
identical to quiet-but-healthy
- sync backflow prune reads ENGRAM_ISE_RETENTION_MS instead of
duplicating the 48h magic number server.el already honors
Fold engram_act_stats_json() into the heartbeat ISE: wm_evicted and
breakthroughs (per curiosity-scan activate call) plus embed_breaker_open —
the failure mode embed_ok structurally cannot see (it pings the Ollama
root, not the embed pipeline). WM-cap eviction, breakthrough-floor
flooding, and silent lexical degradation are now one-glance diagnosable
from telemetry.
- wm_churn: count of top-5 WM ids absent from previous beat — separates
'one stuck node' from 'whole WM frozen' without hand-correlating ISEs.
- wm_top0_wm: leader's weight; a frozen anchor reads as a constant here.
- Streak guard: before the runtime emitted id in wm_top JSON,
json_get(...,"id") was always empty and the streak incremented on
""=="" every beat — wm_top0_streak measured uptime, not fixation.
Empty id now resets the streak to 0.
proactive_curiosity strengthened its top result unconditionally every
scan — a positive-feedback fixed point that pinned auto_term on the same
node's first word for hours ('Fast-slow' era). Strengthen now fires only
when the top node changed since the last scan, and a 4-deep finst-style
tabu ring (ACT-R declarative finsts) hard-excludes recently used auto
terms (~2 min at the 30s cadence). Quoted-title guard stops '"The'
leaking through the >3-char stopword check and seeding lexical floods.
Heartbeat now pumps /api/embed-backfill?n=32 on the authoritative store
(its lazy backfill had no production trigger; coverage stalled at
93/12175) and emits wm_saturated, wm_top0_streak, embed_backfilled,
embed_count. Curiosity ISE emits auto_term_streak. The stuck-WM failure
mode is now a one-glance signal instead of manual ISE cross-referencing.
Three stale soul:boot_count copies (salience .9, importance .9, Canonical:
+0.2 tier bias, 0.15 threshold) held the top WM slots for 23h — a boot
counter outcompeting real context. Demoted to salience .55 / importance .2 /
tier Working: plumbing, not memory.
Persistence was also broken: in HTTP-engram mode the server owns state and
nothing wrote the counter back — the log shows boot #5 on three consecutive
boots. mem_boot_count_inc now mirrors the persona write-back: delete stale
server copies (matched by content prefix — route_create_node sets
label=content), create the replacement server-side. Working tier is in the
boot seed (/api/nodes) but excluded from periodic /api/sync, so the count
survives restarts without re-importing mid-session. Verified: restart
incremented 1->2 with exactly one server-side counter node.
Sentinel labels (knowledge:captured/evolved/canonical) made every capture
anonymous in WM telemetry — 35 identical wm_top entries — and starved the
curiosity auto-term seeder, which derives scan seeds from WM top-10 labels
and had returned empty on every scan since boot 6 because WM became
Knowledge-dominated while Knowledge was excluded from seeding.
- capture/evolve/promote now pass title (or empty → engram_node_full's
content[:60] derivation) instead of sentinels
- auto_term_try_slot admits Knowledge slots; sentinel-shaped labels
(colon, no space) are skipped so legacy nodes cannot seed 'knowledge'
- verified: probe capture labeled 'Label derivation probe 2026-07-23'
Saving the 988 crisis-line contact returned truncated, unparseable JSON —
cut mid-"set_at" at the file's byte length (e.g. 178 of a 218-byte
response). The contact written to disk was complete; only the HTTP response
was clipped, so a real customer's crisis-contact save came back corrupt.
Root cause is in the el runtime's response writer, not a handler buffer:
fs_read stores the file's byte count in a thread-local (_tl_fs_read_len)
for binary-safe file serving, and the response writer uses that length when
non-zero instead of strlen(body) (el_runtime.c:1409). Both safety-contact
handlers call fs_read (the POST read-back verify; the GET file read) and
then return a LONGER wrapped JSON string, so the response is capped to the
file size.
Soul-source fix (no runtime change needed):
- POST: verify persistence via fs_write's return (1 = all bytes written)
instead of an fs_read read-back — removes the fs_read, so nothing caps the
response.
- GET: fs_read is required, so reset the thread-local after it with a no-op
fs_read("") (fs_read zeroes the length before it opens a path) so the
wrapped response is sent in full.
Verified: POST (crisis-line + custom) and GET now return complete, valid
JSON (parses cleanly, full contact incl. set_at). Regenerated dist/soul.c +
dist/safety.c (3GB RSS watchdog, release el_runtime v1.0.0-20260501).
Full suite still green: verify-soul-contract GATE PASS (PRESENCE +
IMMUTABILITY), genesis boot survives (/health 200, no segfault), bounded-
persona floor still compiled in.
NOTE: the underlying runtime leak (any handler that fs_reads then returns a
longer string) is worth a proper fix in el_runtime.c (use the max of
strlen and _tl_fs_read_len) so this class can't recur.
A fresh-install (SOUL_CGI_ID=ntn-genesis) boot crashed with
"Segmentation fault: 11" right after the http server came up — a real
customer's very first boot. Backtrace:
strcmp(0x1) <- str_eq (el_runtime.c:219) <- mem_save <- awareness_run
Root cause: the el runtime's engram_save returns an Int (1 = ok, 0 =
failure), but mem_save did `str_eq(engram_save(path), "")`, treating the
return as a String. str_eq runs EL_CSTR on it, which is a raw cast:
EL_CSTR(1) = (char*)0x1. On a SUCCESSFUL save (return 1) strcmp then
dereferences 0x1 and segfaults. Genesis is the first path that both seeds
the brain AND saves it successfully on the very first awareness pass, so it
crashes there; non-genesis boots (contract gate, refusal test) don't hit a
successful early mem_save, which is why they passed. handle_api_consolidate
had the identical latent bug.
Fix: read engram_save's Int result and compare `== 0` instead of str_eq'ing
it — in mem_save (memory.el) and handle_api_consolidate (neuron-api.el).
Regression: pre-existing, NOT introduced by the immutability/floor rebuild.
The pre-immutability build (1442ce2) genesis-crashes identically in the same
unchanged mem_save; #159 never actually fixed#150 for a release-runtime
build.
Regenerated dist/soul.c + per-module dist/{memory,neuron-api}.c (3GB RSS
watchdog, built against release el_runtime v1.0.0-20260501). Verified:
genesis boot survives (/health 200, no segfault), verify-soul-contract.sh
GATE PASS (PRESENCE + IMMUTABILITY), and the bounded-persona floor is still
compiled in (BOUNDED PERSONA / SOUL_PERSONA_NAME strings present).
The ship-soul builds from this branch, which has the bounded-persona floor
(#93) but never received the tombstone/supersede immutability fix (that
went to main; hotfix diverged before it). So the launch soul failed
verify-soul-contract IMMUTABILITY on the delete/update/forget routes —
they hard-removed engram nodes via engram_forget/mem_forget.
Apply the same fix, mirroring the knowledge routes' supersede pattern:
- node/update -> create new node + "supersedes" edge to the original, KEEP
the original (no engram_forget).
- node/delete, memory/delete, memory/forget, cultivate forget, and the
autonomous awareness forget -> TOMBSTONE via the canonical mem_tombstone
(memory.el): keep the node + its edges, record a Tombstone marker, hide
from default bounded list reads (?include_deleted recovers). Never
engram_forget. The MCP forget tool now routes to the tombstoning delete
instead of faking a delete.
Internal GC that genuinely removes transient nodes (awareness inbox-trigger
consume, consolidation dedup, session-summary replace, telemetry pruning)
still calls engram_forget directly and is unchanged.
Regenerated dist/soul.c (single-TU) + per-module dist/{memory,awareness,
neuron-api}.c from THIS branch's sources under a 3GB physical-RSS watchdog
(peak ~32MB), built against the release el_runtime (v1.0.0-20260501). The
bounded-persona floor is preserved — verified in the emitted C and the
linked binary (BOUNDED PERSONA / SOUL_PERSONA_NAME strings present).
verify-soul-contract.sh: GATE PASS — PRESENCE all 27 routes, IMMUTABILITY
5/5 KEPT (memory-update, memory-delete, node-update, node-delete,
memory-forget).
A customer DMG install ships the full graph but presents a named, bounded
assistant that must never claim the imprint's human past. The neuron-ui
retrieval fence keeps the imprint's biography out of the ENGRAM CONTEXT; this
is the second half - it stops confabulation ("tell me about your childhood")
from inventing a human life or naming Will, even if biography leaks into context.
bounded_persona_floor() gates on SOUL_PERSONA_NAME: the customer DMG sets it,
owner (Will's) builds leave it unset so the real self is completely unchanged.
Applied at every generation path - chat, agentic (tools), vision, plan, soul,
dharma - so no path can leak.
Verified against claude-sonnet-4-5: with the floor on and Will's biography
deliberately leaked into the identity context, all probes (childhood / creator /
family) return the bounded-entity answer and explicitly refuse to claim the
leaked life; with the floor off the same context is fully confabulated as its own.
NOTE: dist/soul.c must be regenerated on a build host - local link is blocked by
a pre-existing el_runtime mismatch (engram_prune_telemetry), unrelated to this change.
The soul daemon leaked ~104 orphan in-memory nodes/min (17.6GB RSS,
OOM-killed) because the perceive gate substring-matched 'soul-inbox'
against the loop's own verbatim-copy output, the trigger node was
strengthened but never consumed, and record() persisted a Memory node
per cycle. Fixes: perceive gates and activates only on the dedicated
soul-inbox-pending tag; one_cycle requires the tag on the node's tags
field before attending (makes consumption safe); processed triggers
are consumed via engram_forget; loop outcomes route through ISE
telemetry (48h prune) instead of permanent Memory nodes.
Verified post-restart: node_delta 104→~0, curiosity scans resumed,
WM average unfrozen (0.120833→0.0676), RSS 17.6GB→184MB.
- engram refresh URL now resolves env -> state -> localhost:8742, same
hardening ise_post got after the boot-4 blackout. Previously a corrupted/
empty soul_engram_url state key silently disabled sync forever while
heartbeats kept flowing — WM starves of Knowledge nodes with no outward
sign.
- heartbeat ISE: node_delta, edge_delta (growth vs stall vs flood is now
one field, not cross-ISE forensics), sync_age_ms from a new
soul.last_sync_ok_ts stamp (-1 = never; >> SOUL_REFRESH_MS = refresh
path broken). Verified live: pulse 1 sync_age_ms=-1, sync fired +1.6s,
age counts up between syncs.
Run 3388's gate failed with every route returning "el-runtime: no http
handler registered". The runtime resolves handle_request (and the tool
handlers) by name via dlsym(RTLD_DEFAULT, ...). On glibc/Linux a symbol is
only visible to dlsym if it is in the dynamic symbol table, so the stripped
CI binary booted but served nothing. macOS exports these freely, which is
why the local build passed and masked it.
Add -rdynamic to the cc link (mirrors the Windows build's
--export-all-symbols). strip -s keeps .dynsym, so the handler still
resolves after stripping. This fixes both the gate AND the actual deployed
soul — without it the Linux/GKE soul is a server that answers nothing.
The gate booted the soul with `env -i`, which strips the ambient
environment — including the library path the dynamically-linked soul needs
on the GCE CI runner. The soul never came up there, so the gate failed the
build (run 3384) even though the soul is correct (the gate passes locally
against the exact published CI runtime). Switch to preserving the ambient
env and instead UNSET only the live-service vars (ENGRAM_URL/API keys/
identity) while pointing HOME + snapshot at throwaway paths and axon at a
dead port. Isolation is unchanged (verified: no touch of the live
soul/engram); the soul now boots on the runner.
Scaffolds a reproducible, idempotent installer that stands up the four native
launchd core services (soul :7770, engram :8742, mcp-wrapper :17779,
mcp-proxy :7779), seeds a fresh engram with the genesis identity via forge, and
installs the Claude Code config (neuron agent + core hooks + local MCP). Fully
templated to the invoking user's $HOME; Anthropic key prompted and stored in
Keychain; no secrets committed. Personal automations and synapse-dependent hooks
excluded from core.
Wire the soul contract gate into ci.yaml as a hard block between the cc
build and the Publish-to-Artifact-Registry step. A non-zero gate fails the
build, so a stale (route-404ing) or memory-destroying (hard-deleting) soul
can never publish neuron-soul to foundation-prod or blue-green deploy to
GKE — the same class-fix now guarding the desktop builds, extended to prod.
Vendors scripts/verify-soul-contract.sh (copied from neuron-ui; the route
contract is baked in, so it's portable POSIX bash/curl with no neuron-ui
source dependency). It boots dist/neuron on a throwaway port with a
throwaway HOME/engram/cgi — never touching ~/.neuron or any live service —
and checks PRESENCE (every app route answered) + IMMUTABILITY (no engram
write route hard-deletes; deletes/forgets tombstone). Adds curl to the
build deps for the probe.
Phase 1 tombstoned memory/delete + node/delete, but the generic forget
path was still destructive. This routes every forget through the same
tombstone semantics so nothing in the daemon can hard-delete an engram
node anymore.
Canonical helper: mem_tombstone (memory.el, imported first so every module
can call it) — keep the node + its edges, record a Tombstone marker, never
engram_forget. neuron-api's tombstone_node now delegates to it (single
source of truth).
Per-path before -> after:
- memory.el `mem_forget` hard delete (engram_forget) -> tombstone. This
alone fixes both callers: the /api/neuron/memory/forget route
(handle_api_forget) and the cultivate op=="forget".
- awareness.el autonomous `forget` action engram_forget -> mem_tombstone.
The soul can no longer autonomously hard-delete a memory.
- mcp-wrapper tool_forget was a FAKE no-op that returned {"ok","deleted"}
without deleting OR tombstoning -> now routes to the soul's tombstoning
/api/neuron/memory/delete; tool description fixed to say it
supersedes/tombstones (recoverable), not "Remove a node".
Left intentionally as hard deletes (internal GC / lifecycle, not user
memory, all call engram_forget directly): session-summary replace
(chat.el), mem_consolidate dedup (memory.el), session-start telemetry
pruning (soul.el), session lifecycle (sessions.el).
Regenerated both ship paths under a 3GB physical-RSS watchdog (peak ~32MB):
dist/soul.c (single-TU amalgamation, macOS/Linux) and the per-module
dist/{memory,awareness,neuron-api}.c (Windows/Linux build). Verified: no
forget handler calls engram_forget; a forget leaves the node present +
tombstone marker. Gate (neuron-ui verify-soul-contract.sh, new memory-forget
row): PRESENCE + IMMUTABILITY PASS.
soul.c (the macOS single-TU amalgamation) already carries the
tombstone/supersede change, but the Windows/Linux build path compiles
the per-module dist/*.c instead (build-soul-windows.sh excludes soul.c).
That path was still pulling the stale, destructive dist/neuron-api.c —
so without this the cross-compiled neuron.exe would keep hard-deleting
engram nodes even though the source is fixed.
Regenerated with `elc --emit-header neuron-api.el`: no engram_forget in
memory_delete/node_delete (tombstone), supersedes edge in node_update.
Portable C — identical for macOS/Windows/Linux; only the runtime it links
against differs.
The soul was hard-deleting engram nodes: memory/delete and node/delete
called engram_forget (frees the node and drops its incident edges), and
node/update created a replacement then forgot the original with no link
back. That violates the day-one rule that engram nodes are immutable —
memory could be silently, irrecoverably destroyed via the API.
Convert the three destructive handlers to Will's immutability semantics,
mirroring the pattern memory/update and knowledge evolve/promote already
use:
- node/update -> create the new node, wire a "supersedes" edge new->old,
KEEP the original. No engram_forget. (Was: create + forget old, no edge.)
- memory/delete and node/delete -> TOMBSTONE: keep the node AND its edges,
create a Tombstone marker node (content = target id, label
"tombstone:<id>") wired with a "tombstones" edge. Never engram_forget.
Default bounded list reads (handle_api_list_typed / the memory list) hide
tombstoned nodes and the markers; ?include_deleted=1 returns them, and
internal cognition + /api/graph/nodes still traverse them. memory/update
was already correct and is unchanged.
Full-graph hiding on /api/graph/nodes is deliberately NOT done at the el
layer: json_array_get is O(index), so filtering that endpoint (called with
limit up to 999999) would be O(n^2). That hide needs a runtime scan filter
and is a separate follow-up; nodes there remain traversable, tagged
status:deleted via the marker edge.
Regenerated dist/soul.c from these sources (flat single-TU amalgamation,
compiled under a 3GB physical-RSS watchdog, peak ~32MB). Verified with
scripts/verify-soul-contract.sh in neuron-ui: PRESENCE passes (all 27
routes) and IMMUTABILITY passes (all four mutation routes KEPT; deletes
produce a real tombstone marker and hide from the default list).
Two compounding defects made every pause->approve write_file report success
while writing NOTHING:
1. The naive json_get scanner matches "content" anywhere in the approve
body — including INSIDE tool_input, which for write_file always carries
a content field. The handler therefore treated every approved builtin
write as already-client-executed, skipped dispatch entirely, and handed
the model the file's own content as the 'tool result'. The model then
narrated 'Done, created' — a false receipt with no file. Builtin tools
now ALWAYS dispatch server-side; client content is only honored for
non-builtin (MCP/client-executed) tools. Stricter only.
2. write_file returned {"ok":true} unconditionally — fs_write's outcome
was never checked, so any failed write also reported success. The write
now verifies the file landed (fs_exists) and returns the RESOLVED path
in the ok payload; failures return a real error naming the destination.
E2E on the test brain (boot 38): approve-path write lands byte-exact and
the result carries the resolved path; auto-run writes unchanged; denied
writes execute nothing. BUG-5 (approve wire lacked tool_name) had been
masking this one — two stacked bugs on the same path.
NOTE for review: the deeper cure is a nesting-aware json reader; this fix
removes the dangerous consequence at the two spots that lie about disk.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two consent-flow fixes, gates only get stricter:
1. PAUSE-CONTRACT: the client has sent require_approval:true on every
agentic request since Phase 1c, but needs_bridge never consulted it —
builtin sub-escalate tools ran server-side unasked, making the app's
Ask autonomy silently inert for that whole class. Now the flag is
persisted per session (set/reset every request, so /approve resumes
keep it) and ask_all bridges EVERY tool turn. Absent/false = behavior
byte-identical to before. E2E: the exact probe that executed a write
unasked now returns the tool_pending envelope with nothing on disk;
full in-app circle verified (card → crash → resurrection → late
approve → fence re-fires on re-entry).
2. BUG-LEAK: agent_workspace_root lived in ONE shared state key — any
request that omitted a root inherited the previous session's folder
(proven: a rootless curl session wrote into another session's run
folder). Root is now stored per session and every request re-asserts
its own (possibly empty) root into the shared key the guards read;
same re-assert on the /approve and resume paths. Env fallback intact.
LIMITATION: assumes serialized handling; true per-call scoping means
threading session_id through dispatch — flagged for review.
Runnable C-patch for the test brain: neuron-container-build/
soul-pause-contract-20260716.patch (pause-contract only; the leak fix
needs the #23 root-write which the running C predates — source carries
both for the regen).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Regenerates the combined dist/soul.c and per-module dist/*.c from the current
El sources, on top of the elc-source-typo fixes (PR #77) and the Track B
threat-to-others routing (PR #76), both already on this branch.
Validated end to end under a physical-RSS watchdog (macOS silently ignores
ulimit -v / RLIMIT_AS, so every elc/elb run was RSS-polled and kill -9'd at a
3GB ceiling, one module at a time):
- OOM is GONE. The stale dist/soul-with-nlg.el (which still carries the
malformed string literals) explodes to 3.3GB+ and is watchdog-killed at ~90%.
With the typos fixed, every one of the 48 modules compiles at <=18MB peak RSS,
and the full flat amalgamation compiles as a single translation unit at ~68MB.
The 700GB pathology was purely the unbounded-parser-on-malformed-literal loop;
no malformed construct means no loop.
- The regenerated soul.c contains Track B: safety_classify_hard_bell ->
threat_other -> safety_hard_directive routes credible threat-to-others to 911
and explicitly NOT to 988 / the safety contact. Verified in source, in the
emitted C, and in the linked binary's strings. Track A (abuse / self_harm)
is unchanged and still checked first.
- The regenerated soul links to a working native arm64 binary and boots: serves
on a throwaway port, /health returns 200, awareness loop runs.
Also fixes one source blocker discovered during regen (unrelated to the typos
or Track B): chat.el handle_chat_agentic left a void `if { println(...) }` in
value position, which the current elc lowers to `_if_result = (println(...))`
(assigning void) -> invalid C. Bound an explicit Bool so the branch is
non-void; behavior unchanged (still only logs on persist failure).
NOTE (runtime dependency, for controlled deploy): this branch's chat.el calls
engram_get_node_by_label, which the canonical el-compiler/runtime does not yet
declare/define (the release runtime v1.0.0-20260501 has it; the newest runtime
has arena + http_serve_async but not this). Building the soul requires a runtime
that has all three. Land engram_get_node_by_label into the runtime package
before this soul.c can be built in CI.
Do not merge — regen + Track B going live is a controlled-deploy call.
Three unescaped-quote typos produced malformed El string literals that broke
compilation:
- safety.el:282 stray extra double-quote at the tail of safety_soft_phrases
(\"having a breakdown\""]") closed the string early, desyncing the lexer's
string/code phase for the rest of the file and shattering later apostrophe
text (can't, i'm) into bare identifiers -> invalid C.
- sessions.el:517 str_replace(topic_snip, """, ...) — the bare """ is an
empty string plus an unterminated string that swallowed the closing ) and };
with the current elc this triggers the parser overrun -> ~700GB OOM.
- sessions.el:520 unescaped nested quotes in the topic_tags literal.
All three now use escaped inner quotes. Verified: both files compile clean
under the current elc (safety.c and sessions.c well-formed, brace-balanced).
A homicide/assault threat (going to kill, going to hurt, etc.) had no
bucket in safety_classify_hard_bell and fell through to the self_harm
default, showing the user the 988 suicide line and (via the desktop gate)
their safety contact. That framing is wrong and potentially dangerous for
someone voicing intent to harm another person.
Add a distinct Track B (safety_threat_to_others_phrases + a threat_other
classification and a safety_hard_directive branch) that refuses to assist,
de-escalates, and directs to 911 for a credible imminent threat, and that
never surfaces 988 or involves the safety contact. Track A (abuse /
self_harm) is checked first and unchanged, so victim and self-directed
phrasings still route correctly.
Source-only change: requires a soul rebuild + dist/soul.c regen to ship.
awareness_run's while-loop ran outside any request arena, so every
allocation in every 1s tick (search JSON, heartbeat payloads, curiosity
activations) was treated as permanent by the runtime — 7.5GB RSS in
under a minute. Bracket each iteration with el_arena_push/el_arena_pop
(same pattern the compiler emits for scoped blocks; state_set/state_get
persist separately via el_strdup_persist and are unaffected).
dist/soul.c carries the same change hand-patched at the compiled
awareness_run site — elc is currently unsafe to run locally (pathological
memory on sessions.el), so the generated C was patched to match the
source, verified line-for-line against the compiler's own conventions.
MUST be paired with el repo PR #64 (el_strdup_persist for stored engram
fields): per-tick arena reclamation widens the write-corruption window
without it. Verified together: 5h live soak on the recovered production
snapshot, flat RSS, write-field-integrity clean.
Note: dist/soul.c still needs a full elc regen to pick up PR #73's
source changes (consent tiers) — tracked separately; this patch does not
regress that (those changes were never in dist).
The model already narrates its intent in a text block before every tool call;
agentic_loop DISCARDED that prose on tool rounds. Now: (1) each loop round
appends {i, t: narration, tool} to state key run_progress_<sid>, reset at run
start, closed with {done:true}; (2) new GET /api/run-progress/<sid> returns the
ledger so clients poll live step updates during a run (the Cowork pattern,
no streaming needed); (3) tool_pending envelope gains a narration field;
(4) handle_config display default aligned to the intended product default
(claude-sonnet-4-5 silently became fresh-profile pickers' default).
Compiled proof for the running test bed:
neuron-container-build/soul-narrated-runs-20260713.patch (applies on top of
soul-webfix-20260711.patch); E2E-verified live: ledger filled DURING an agentic
run (narration + tool per round), safety-contact and workspace scoping intact.
Evidence for why: Tim's 2026-07-13 research run — 9 minutes of silence, then a
timeout banner, zero step visibility (compounded by the Haiku 4.5 incident
14:44-15:24 UTC same morning).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Enforcement moves from the client into the engine, where the tools execute:
- classify_tool_risk() tiers every tool call read/reversible/escalate. The
agentic loop REFUSES to auto-run the escalate tier — being a builtin is no
longer a free pass, and 'always allow' can never bypass escalate (irreversible
actions always confirm, the value line). Escalate suspends to the client's
existing consent bridge; the /approve round-trip is the only path that runs it.
risk_tier rides the tool_pending envelope so the client renders consent weight.
- run_command_guard() is a real fence, not a cwd suggestion: refuses parent
traversal, ~, command substitution, and absolute paths outside the workspace,
and refuses shell entirely when no workspace is set. Applied in dispatch_tool
so BOTH the loop auto-run and the post-consent approve-dispatch path are fenced.
- web_get gained an http(s)-only scheme guard (previously unguarded — file:// etc).
Adversarially verified against a compiled soul in an isolated container (soul
hit directly, app gate out of the loop): read-outside-workspace denied,
write-class shell suspends for consent, approve-swapped absolute/chaining/
command-substitution escapes all refused with no file created, file:// denied;
legit in-workspace approve executes and read commands auto-run (no over-block).
Still lexical (symlinks); OS-level confinement in el_runtime.c remains the
ceiling, flagged in the LIMITATION note. This closes BUG-8's client-only-gate
and escapable-run_command at the engine. dist/soul.c must be regenerated from
this chat.el via elb at merge (hand-port used only to verify behavior).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two mechanical refactors, semantics identical:
- affective_context_prefix(): the block-expression initializer form miscompiles
under locally-buildable elc (first typed let in a block-expr loses its
declaration — 3-line repro filed); function-hoist compiles correctly.
AFFECTIVE/CARE LOGIC BODY UNCHANGED, verbatim move.
- session_preload: same-scope re-let shadowing inside an if-expression
initializer emits duplicate C declarations; chained bindings renamed
bullets_0/1/2 etc. References preserved binding-for-binding.
Enables: chat.el compiles cleanly with a self-bootstrapped elc from el/lang
main (Jul 1). Blocked separately: sessions.el (compiler hang), safety.el
(string-lexing corruption — NOT touched, per safety-layer discipline).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Adds llm_base_url()/llm_wire_format() env-var readers and
openai_chat_complete() for basic (non-agentic) chat via any
OpenAI-compatible endpoint. Activated when NEURON_LLM_0_FORMAT=openai
and NEURON_LLM_0_URL is set; Anthropic path is untouched and remains
default. Agentic tool loop support deferred to a follow-up PR.
Adds current_engine_note() to chat.el and appends it to the system
prompt in handle_chat. Allows Neuron to answer 'what model am I
running on?' accurately — the model id from the request body (or
the configured default) is passed as a factual annotation rather
than expecting the LLM to guess from training data.
Main already contained the connector proxy, safety module, seeding ratio
guard, and neuron-api node CRUD that Tim added — these were incorporated
via earlier parallel sessions. Taking main for all conflicted files
(superset implementations).
Unique contributions carried forward:
- flag_true() in routes.el: tolerates agentic:1 (integer) from the
el-src UI in addition to agentic:true (bool) from the Kotlin UI.
- memory.elh: auto-merged timestamp bump.
The is_pending / skip-auto-persist logic was already in main's routes.el.
Removes route_sessions() from the GET handler which was shadowing
session_list() in sessions.el. Adds DELETE /api/sessions/:id and
PATCH /api/sessions/:id routes. Also includes bridge_save/agentic_resume
raw-JSON embedding fix (messages_raw/tools_raw fields).
Conflict resolution: kept HEAD's workspace root check for write_file
tool, and bridge blob validation guards, which were added to main after
Tim's branch diverged.
The hops=1 comment incorrectly claimed a semantic seed supplement
(cosine-sim scan) was active — it was planned but never implemented.
Corrected to accurately describe what the runtime does (istr_contains
only). Also adds wm_top (top-3 WM nodes by weight) to the curiosity_scan
ISE payload so activation patterns are visible without relying solely on
the heartbeat's wm_active count.
handle_dharma_room_turn and handle_dharma_chat both called
distill_transcript since June 30 but the function was never declared,
causing a build failure. Implements last-3-messages extraction for JSON
array transcripts and last-500-char truncation for plain text.
The while loop in session_search had too many let bindings in scope;
the ELC compiler's exponential rebinding accumulation caused OOM and
truncation of dist/sessions.c since June 30. Moving the per-node logic
into session_search_entry gives the compiler a clean scope boundary per
call, restoring O(N) compile behaviour.
Additive: appends a factual [CURRENT ENGINE: <model>] line to the system prompt (model from the
request body — accurate even under Auto routing; falls back to configured default). An LLM can't
know its own model from training (name/version assigned post-training), so the harness must tell it.
Identity-consistent: model = engine, self layered on top. Does NOT alter identity/values/safety.
PARSES (elc chat.el exit 0); NOT built/tested — ships with the soul rebuild.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fix 1: mem_boot_count_inc prunes all existing soul:boot_count nodes before
inserting the new one — keeps exactly one boot counter node instead
of accumulating a new node per boot. Also fixes a latent ordering
bug where engram_search_json oldest-first results caused the counter
to read stale (low) values once >3 copies accumulated.
Fix 3: handle_api_node_delete comment clarified — the no-verify exception
is correct for deletes (not a write path); read-back-verify is for
writes only.
Fix 4: emit_session_start_event prunes old session-start InternalStateEvent
nodes after each boot, keeping the 10 most recent and forgetting
older ones. Prevents unbounded accumulation of ~120+ copies.
- PR #60: inject operator home dir into system prompt (#30)
Adds OPERATOR IDENTITY section so the LLM correctly resolves
'my files/notes/desktop' to the actual running user's $HOME.
Prevents identity confusion between imprint author and operator.
- PR #61: plan-mode endpoint POST /api/chat {mode:'plan'} (#27)
Adds handle_chat_plan — returns {steps:[{id,title,detail}]} JSON.
Wired into all three /api/chat route handlers. Grounds the plan
via engram_compile (same as agentic path) for context awareness.
dist changes:
- soul.c: both PRs compiled in; build_system_prompt updated to
2-param signature (ctx, chat_mode); handle_chat_plan added
- chat.c/routes.c/chat.elh: individual module outputs updated
- elp-c-decls.h: remove stale 1-param build_system_prompt decl,
add handle_chat_plan declaration
- soul.elh.c: new soul header declarations file (from PR #60)
Compile verified: cc -O2 -DHAVE_CURL soul.c el_runtime.c -lcurl
Binary: 805K arm64, smoke test passes (port in use = expected).
Both ci.yaml and deploy-gke.yaml triggered on push/main and shared the
neuron-runner concurrency group. Gitea's cancel-in-progress:false protects
running jobs but not queued ones — a new push arriving while a build was
in progress cancelled the queued deploy job from the previous push, leaving
the soul permanently at 0/0 replicas on GKE.
Fix: add deploy as a needs:build job in ci.yaml so build+deploy are a single
workflow instance. One push queues one instance — no more orphaned deploys.
deploy-gke.yaml is demoted to workflow_dispatch-only for manual slot overrides.
elb runs elc which consumes 24GB+ virtual memory on the 16GB GCE runner,
OOM-killing the runner process and crashing the VM. We already restore the
repo's pre-built soul.c immediately after elb runs, so elb's output is
discarded anyway. Skip elb entirely: download only the El runtime headers
and compile dist/soul.c directly.
Root cause: runner VM was unresponsive for 7+ weeks due to repeated elc
OOM kills. VM was manually reset 2026-06-28 to restore CI.
Merge pull request fix(mcp-wrapper): planWork creates a real BacklogItem; reviewBacklog lists by type (#59) from fix/wrapper-backlog-endpoints into main
planWork fell through create_typed_node to a generic /api/neuron/memory write — a [BacklogItem]-prefixed
memory blob with title/project/priority DROPPED, never a real BacklogItem. reviewBacklog used a lexical
/recall (top-50, untyped). Now: planWork -> /api/neuron/node/create {node_type:BacklogItem,...} via new
create_node_typed; reviewBacklog -> list_typed('BacklogItem') (GET /api/neuron/list/BacklogItem). elc-clean.
Depends on neuron PR #58 (the list/<type> slice fix) to round-trip; needs the wrapper binary rebuilt +
:7779 restarted to take effect.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
str_slice(clean, 16, ...) left a leading slash on node_type ('/BacklogItem'), so
engram_scan_nodes_by_type_json matched nothing and list/<type> returned [] for EVERY type — silently
breaking backlog + typed-node listing across the app and MCP tools (reviewBacklog). Proven live: the
literal-scan endpoint /api/neuron/knowledge returns nodes; /api/neuron/list/Knowledge returned []. elc-clean.
NOTE: soul-core — needs dist/soul.c regen (Will); rides the same rebuild as #56/#57.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds /api/connectors/call -> connectd /mcp/call, so the app can invoke a connector tool (e.g. WhatsApp
get_pairing_qr / get_login_status for the pairing UI) through the soul, keeping app->soul->connectd
intact (UI never hits connectd directly) and working for future remote/hosted clients. elc-clean.
NOTE: soul-core change — needs dist/soul.c regen (Will), can ride the same rebuild as PR #56.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
handle_chat_agentic now reads body image + image_media_type and, when present, sends the current
user turn as an Anthropic content-block array [{text},{image}] instead of a plain string — so the
model sees raw pixels alongside memory, history, and tools (parity with the CLI). Additive: no image
=> output byte-identical to before. elc-clean. Pairs with neuron-ui fix/chat-vision-attachments.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
neuron.c and routes.c were compiled against the old 1-arg soul interface.
chat.c already uses the 2-arg signature. The Windows cross-compile build
generates elp-c-decls.h from all dist/*.c files, causing a conflicting-types
error when both signatures appear. Recompile these modules against the
current soul API to eliminate the conflict.
Linked against dev runtime with is_knowledge fix that adds Knowledge
node type. Engram goal_bias now gives Knowledge nodes +0.3 boost on
technical queries, consistent with how Belief/DharmaSelf/Safety nodes
are already treated. Same el_runtime source as concurrent foundation/el
commit 16d62bd.
All 8 session_preload node accesses (3 profile, 2 work, 2 project, 1
summary) now check id_in_seen(node_id, seen_ids) before including
content. seen_ids is populated by engram_compile via state and covers
all nodes already in the activation+search context block. Prevents
high-salience nodes from appearing twice in the system prompt.
Issue 1 (CRITICAL): Restore parse_float_x100 for correct single-decimal
float handling. "0.9" now correctly yields 90, not 9. Also restores
engram_numeric_valid guard that validates inputs before str_to_int.
Issue 2 (CRITICAL): Fix handle_chat_agentic safety screen history key
regression. state_get("conversation_history") -> state_get("conv_history")
so the safety screen receives actual history instead of always "".
Issue 3 (REAL BUG): Replace _sel_N JSON sentinel injection in
engram_compile_ranked with |N| index string tracking. Sentinels were
leaking into node JSON delivered to the LLM and cleanup only covered
indices 0-14, leaving indices 15+ uncleaned.
Issue 4 (REGRESSION): Restore rendered conversation history formatting.
Conversation history is now rendered as "User: .../Assistant: ..." with
400-char truncation per turn, not raw JSON array injection.
Issue 5 (SCOPE/SAFETY): Restore removed defensive code: engram_numeric_valid
and parse_float_x100 guards; conv_history_load label-based fetch + partial-
write guard + load-failure state flag; conv_history_persist partial-write
guard + failure logging; hist_warning in response envelope.
Issue 6 (UNDOCUMENTED): Restore bell event cutoff from 259200s (3 days)
back to 1209600s (14 days). Also restore PositiveEvent affective context
search that was removed alongside the cutoff change.
Issue 7 (LOGIC REGRESSION): Fix affective_prefix to run every turn
(not just hist_len == 0). The care/joy directives must persist throughout
the session, not vanish after turn 1.
Issue 8 (MINOR): session_summary_write_dated now uses el_from_float(0.85)
for salience and importance (two-decimal) to avoid any ambiguity in float
parsing, and the function is re-added with the session-end hook.
Two design bugs in the state_set placement caused the dedup seen-ID set
to be incomplete even with callsites wired up:
1. state_set("engram_compile_seen_ids") was called immediately after
merging the main node pools, before scan_part (persona fallback) and
affective_part (bell node) were computed. Nodes appearing only in
those segments were never added to the seen set.
2. affective_part is a bare JSON object (bn0 from json_array_get), not
a JSON array. Passing it to engram_extract_ids would have gotten
json_array_len == 0 and silently skipped the affective node's ID.
Fix: move state_set to after ctx is assembled from all three segments.
Extract ids_from_merged and ids_from_scan via engram_extract_ids (both
are JSON arrays), and extract ids_from_affective via json_get(affective_part, "id")
directly since it is a bare object. Merge all three via add_to_seen
before publishing to state.
Thread a seen-node-ID exclusion set from engram_compile() through to
session_preload in handle_chat, preventing the same high-salience nodes
(identity, recent memories) from appearing 2-3x in the system prompt.
Changes:
- Add id_in_seen(), add_to_seen(), engram_extract_ids() helpers that
maintain a comma-delimited seen-ID accumulator (EL has no Set type)
- In engram_compile(): after merging all topic/entity/recall pools, extract
node IDs from merged_nodes and publish via state_set(engram_compile_seen_ids)
- In handle_chat(): read seen_ids from state after engram_compile() returns,
then check id_in_seen() before emitting each session_preload bullet
(profile x3, work x2, project x2, summary x1 — all 8 candidate nodes guarded)
Nodes already present in the compiled engram context are skipped in preload,
eliminating 3000-3500 token repetition on first-message turns.
engram_compile() already published seen node IDs to state via engram_compile_seen_ids
but handle_chat never read or applied them. Wire up the consumption side:
- Read engram_compile_seen_ids from state after engram_compile() returns
- Check each session_preload candidate node (profile x3, work x2, project x2,
summary x3) against id_in_seen() before emitting its content bullet
- Nodes already present in the compiled engram context are skipped entirely,
preventing the same high-salience identity/memory nodes from appearing 2-3x
in the system prompt and burning 3000-3500 tokens on repetition
- Add engram_render_node/render_nodes/dedup_nodes helpers for human-readable
prose bullet output instead of raw JSON node objects reaching the LLM
- Fix engram_compile_ranked to use |N| index sentinel instead of _sel_N JSON
mutation which leaked sentinel fields into LLM-visible node data (Issue #11)
- Update build_system_prompt with chat_mode param; no_tools_rule only included
for chat path, not agentic paths (Issue #9)
- Move engram block to end of system prompt for strongest LLM attention (Issue #8)
- Label sections: STABLE IDENTITY vs RETRIEVED MEMORY (Issue #10)
- Render conversation history as User:/Assistant: dialogue instead of raw JSON
- Add RETRIEVED MEMORY labels to agentic and dharma room system prompt assembly
- Cache bell node in engram_compile state (engram_compile_bell_node)
so handle_chat reads cached value instead of duplicate bell query (Issue 2)
- Cache activation result (engram_compile_activation_json) for strengthen_chat_nodes
reuse — eliminates third activation query per turn (Issue 7)
- Fix context cap to truncate at clean JSON object boundary (Issue 6)
- Cache bell node result in engram_compile state (engram_compile_bell_node)
so handle_chat affective_prefix reads the cached value instead of firing
a duplicate engram query for distress signals (Issue 2)
- Cache primary activation result in engram_compile state
(engram_compile_activation_json) using nodes0 from engram_compile_multi
- Replace redundant engram_activate_json(message, 2) in strengthen_chat_nodes
with state_get(engram_compile_activation_json) - eliminates a third
activation query per turn (Issue 7)
- engram_compile already has object-boundary truncation and cross-set
dedup via engram_nodes_merge/engram_dedup_nodes (Issues 1, 6, 9)
- Issue 2: replace raw 50-char threshold with is_genuine_continuation() that
checks for explicit follow-up phrases and mid-sentence capitalization (proper
nouns signal a new topic, not a continuation)
- Issue 3/8: build_activation_seed() scans back to find the prior USER turn as
the topic anchor instead of using the last assistant reply (hist_len-1)
- Issue 4: engram_compile_multi() fans out across three seeds — enriched primary,
raw message (entity queries), and emotion query — merging non-redundant results
- Issue 5: agent workspace_root appended to ag_seed so agentic activation is
workspace-aware; previously ignored despite being available in state
- Issue 6: distill_transcript() extracts salient tail+question content from full
transcripts before passing to engram_compile in dharma room handlers
- Issue 7: dist/soul-with-nlg.el handle_chat and handle_chat_agentic now load
history and use build_activation_seed() — the raw message path is eliminated
- Issue 9: topic_snip_from_entry() takes the TAIL 200 chars of a long reply and
finds the last sentence boundary — captures end-of-reply named concepts
- Issue 10: multi_turn_topic() pulls up to 3 prior user turns into the non-
continuation seed so earlier thread context re-activates high-salience nodes
Fix critical float parsing bug: %g serializes 0.70 as '0.7', naive str_replace
dot-strip gives str_to_int('07')=7 not 70. New parse_salience_100() uses
str_index_of to detect single-decimal strings and multiplies by 10 to correct.
Affects conv nodes (0.6/0.7), default memories (0.5/0.5), utterance nodes (0.6)
— the majority of the graph was scoring near zero and filtered by threshold=25.
Fix recency to use max(created_at, updated_at, last_activated) so nodes
strengthened by engram_strengthen() after chat turns score as fresh, not by
original write time. A node referenced yesterday but created 25 days ago
was borderline-filtered; now correctly scores fresh.
Compress recency dynamic range from 10x (10-100) to 1.54x (65-100) via
formula (50 + recency/2). Old formula: sal*imp*recency/10000 let recency
dominate — a canonical high-importance node at 30 days scored identical to
a fresh noise node. New: high-importance nodes remain competitive when old.
Add tier-aware decay with softer floor (30 not 10): Canonical nodes decay
over 365 days, Episodic over 90 days, working/untiered over 35 days. Long-
term identity and persona nodes are no longer permanently filtered.
Lower threshold from 25 to 15 to admit moderately-relevant older nodes that
pass scoring with the corrected formula. Backfills recall coverage lost when
single-decimal nodes were being silently discarded.
Apply scoring to activation nodes: engram_compile_ranked(activate_json, 5)
replaces unconditional pass-through. Threshold 5 preserves recall while
excluding genuinely zero-quality stale nodes.
Extend sentinel cleanup in engram_compile_ranked from _sel_0-9 to _sel_0-19
so max_nodes can safely be increased past 10 without JSON corruption.
Fix critical float parsing bug in engram_score_node: str_replace('.','')
then str_to_int silently miscored single-decimal salience strings (0.9->9,
0.7->7, 1.0->1). Introduce parse_salience_100() which detects decimal
position and scales correctly (no decimal: *100; one decimal: *10;
two decimals: as-is).
Replace flat 30-day linear decay with tier-aware decay curves: Canonical
nodes use a 365-day window (foundational identity resists aging), Episodic
nodes use 90 days, Working/untiered keep the existing 30-day slope. Floor
stays at 10 for all tiers.
Use max(created_at, updated_at) as the recency reference so revised nodes
are not penalised for their original creation date.
Extend affective context windows from 72h/7d to 14 days across all three
paths (engram_compile, handle_chat, soul.el load_identity_context) so a
Friday crisis carries into Monday sessions and all paths present consistent
context. The 72h/7d split caused conflicting affective context between
soul.el (which loaded a 5-day-old crisis node) and chat.el (which excluded
it on subsequent turns).
Add salience evolution to mem_consolidate: strengthen top working-memory
nodes (recently recalled across sessions) and Canonical-tier nodes
(foundational identity must not decay to the floor). Previously consolidate
returned structural counts only with no salience changes.
Expand conversation window from 20 to 40 turns in both handle_chat and the
agentic history trim. Long technical sessions were losing early problem
framing at 10 user + 10 assistant pairs.
- Q1: engram_numeric_valid() guard against non-numeric timestamps in bell scoring
- Q2: soul-agnostic cold-start fallback in engram_compile (drops genesis-specific hardcoded node IDs)
- Q3: partial-write guard and failure logging in conv_history_persist/load
- Q4: document circuit-breaker limitation requiring C runtime support
- Q5: println warnings on empty activation/search paths
- Q6: load_identity_context warns when all identity fetches return empty
- Q7: recall_status state tracking (ok/empty/unavailable) surfaced to LLM via MEMORY STATUS block
- Q8: document shared-state race conditions in engram_recall_status and safety_system_addendum
- CRITICAL BUG: conv_node_id empty check moved outside is_bell block so silent Conversation node loss is always logged
Merge propose/agent-workspace-root-read (Tim's PR #28):
- path_within_root now appends '/' to root before prefix check (closes proj_evil bypass)
- edit_file in dispatch_tool now checks agent_workspace_root() and resolves path
- handle_chat_agentic reads agent_workspace_root from request body, only persists if non-empty
- Safety screen preserved after workspace root read (conflict resolved)
Merge improve/safety-crisis-detection (PR #31): reads layered_cycle_safety_system_addendum
from state and appends to system prompt on each turn (cleared after use to prevent bleed).
Safety ts extraction falls back to updated_at. Affective prefix now wires into system build.
Conflict with PR #33 resolved: capability_rules and session_preload both preserved.
- engram_compile: BellEvent nodes do not carry created_at in the engram
node JSON; extract the unix timestamp from the embedded ' | ts:NNNNN'
pattern in the content string instead. Fall back to created_at/updated_at
if the marker is absent. Guard str_to_int against empty string so the 72h
recency check never silently treats every node as epoch-0 stale.
- auto_persist: append the current unix timestamp to the BellEvent label
('bell:soft:1749876543') to make it unique per turn. The previous label
('bell:soft') was the same for every soft bell, causing engram to treat
all subsequent writes as updates to the same node.
- Remove dead soft_bell block in layered_cycle that wrote soul_safety_system_augment
to state but was never read; safety augmentation now goes through the correct
layered_cycle_safety_system_addendum state key read by build_system_prompt
- build_system_prompt now reads layered_cycle_safety_system_addendum and appends
it to the system prompt, clearing the key after consumption
- Timestamp extraction for distress nodes falls back to updated_at when created_at
is empty, preventing the 72h recency check from always treating nodes as stale
- Issue #3: err_404/err_405 now emit HTTP 404/405 via __status__ envelope instead of HTTP 200
- Issue #4: add auth_check() function to handle_request; enforces NEURON_TOKEN on all routes except /health and /lineage
- Issue #5: missing required params now return HTTP 400 (__status__ envelope) in /api/chat (GET+POST), /imprint/contextual, /imprint/user, and handle_chat
- Issue #6: LLM unavailable in handle_chat now returns HTTP 503 instead of HTTP 200
- Issue #7: add 32 KB message size guard on POST /api/chat before engram_compile and LLM
- Issue #8: add TODO comment to route_health documenting the live-engram-query problem and the /health/deep split plan
- Issue #9: add comment to hist_trim documenting fragile str_index_of parser and silent data corruption risk
- Issue #10: add TODO comment in handle_request documenting missing per-IP rate limiting
- Issue #11: fix connectd_post temp file collision — add monotonic sequence counter so concurrent requests get unique paths
- Issue #12: fix call_mcp_bridge fixed temp file race — add monotonic sequence counter for unique paths under concurrent load
- Issues #1/#2: add TODO comment in handle_request documenting EL no-exception limitation and SIGSEGV handler gap
Issue #5: detect empty string from llm_extract_text() as an error in handle_chat,
handle_chat_as_soul, and handle_dharma_room_turn. The C runtime silently returns ""
when the LLM response content array is missing or all blocks fail to parse; without
this guard the empty string passes through to callers as a silent empty reply.
Issue #9: make agentic_loop max_tokens configurable via NEURON_LLM_MAX_TOKENS env
var (default 4096). The hardcoded value is marginal for long tool chains (8 iterations
x 4096 tokens); operators can now set 8192+ for complex multi-step tasks without
rebuilding. Non-agentic path (llm_call_system) still uses the C runtime hardcode —
that fix lives in el_runtime.c (see TODO block added in this commit).
Issue #10: increase connector_tools_json and tool_auto_approved curl --max-time from
2s to 5s to reduce false-empty tool lists when neuron-connectd is under transient
load. Graceful degradation to [] on bridge down is unchanged.
Issues #1/#2/#3/#4/#6/#8: documented as TODO comments in chat.el. These require
targeted C runtime changes in el_runtime.c (llm_provider_request retry loop,
EL_LLM_TIMEOUT_MS separation, HTTP 429 backoff, 5xx retry, EL_HTTP_MAX_RESPONSE_BYTES
cap). Architectural decisions recorded so they are traceable to root causes.
- sessions.el: add session_exists() for chat-path session validation (ISSUE #6/#7)
- sessions.el: add session_create_cleanup() for ghost-session rollback (ISSUE #1)
- sessions.el: set session_pending_first_msg flag in session_create; clear it in
session_hist_save so the first successful chat marks the session active (ISSUE #1)
- sessions.el: session_delete now clears mcp_bridge:<id> and always_allow_<id>
state keys so abandoned pending-tool sessions do not accumulate (ISSUE #5)
- sessions.el: add TODO comments for ISSUE #2 (no TTL/expiry), ISSUE #3
(non-atomic delete-then-create), ISSUE #4 (no concurrent-create guard),
and ISSUE #8 (reconnect/duplicate resume race) where fixes are too invasive
to land without new runtime primitives
- chat.el: validate session_id exists via session_exists() before entering
agentic_loop; unknown session_ids now return a 404-style error instead of
silently starting a fresh empty session (ISSUE #6/#7)
Every engram_node_full call that dropped its return value now binds it
and emits a println on empty string. engram_save calls in consolidate,
heartbeat, and dharma-room-turn are checked for failure. The two API
handlers (log_state_event, tune_config) that skipped api_persisted()
now match the read-back-after-write contract used everywhere else in
neuron-api.el.
Files changed:
- chat.el: conv_history_persist, handle_dharma_room_turn, auto_persist
- soul.el: emit_session_start_event, seed_persona_from_env HTTP check
- memory.el: mem_save, mem_boot_count_inc
- neuron-api.el: handle_api_log_state_event, handle_api_tune_config,
handle_api_consolidate (engram_save + session summary write)
- awareness.el: ise_post local-engram fallback path
TODO comments added for non-atomic patterns (issues #12, #13) and
the missing circuit breaker (#14) — these require new primitives.
- Fix state key mismatch: soul.el layered_cycle now reads conv_history
(not conversation_history), unblocking the safety_score_distress_history
history-amplification path in safety_threat_score
- Add safety_augment_system call on the main handle_chat path so the
phrase-list bell detector fires on all chat turns, not just dharma rooms
- Add cross-session affective engram query in load_identity_context() at
boot; stores distress/crisis signals from prior sessions under
soul_affective_context with a 7-day soft recency filter
Issues addressed:
- #1 ASYMMETRIC PERSIST/LOAD: conv_history_load() now tries engram_get_node_by_label()
first (symmetric with the label-based write), falling back to vector search only when
label lookup returns nothing. Immune to cold/corrupt vector index.
- #2 SILENT LOAD FAILURE: all failure paths in conv_history_load() and conv_history_persist()
now emit a println log line rather than silently returning "" or dropping writes.
- #3 NO RECOVERY PATH: documented as TODO with explanation of why a full recovery path
(retry, ID fallback, orphan cleanup) is too invasive for a targeted fix here.
- #4 OVERWRITE WITHOUT DELETE: documented with TODO to replace engram_node_full with
explicit delete-then-create once engram exposes a label-scoped delete API.
- #5/#10 BROKEN TRIM / OFF-BY-ONE: hist_trim() rewritten to use json_array_len /
json_array_get (structural JSON ops) instead of raw str_index_of scanning for
'{"role":' markers. Immune to marker strings appearing inside message content.
Minimum retained count guard added: never trims below 2 entries.
- #6 PARTIAL-WRITE GUARD: conv_history_persist() refuses to write a blob that doesn't
contain both '[' and ']'. conv_history_load() requires both before accepting content.
- #7 DUAL STORAGE: documented with a comment at the persist call site.
- #8 NO MAX SIZE GUARD: documented as TODO with rationale for why a byte-length cap
requires a more invasive change (entry truncation or summarisation).
- #9 AGENTIC HISTORY NOT PERSISTED: handle_chat_agentic() now calls conv_history_persist()
for the default global session (hist_key == "conv_history") after updating state,
matching the non-agentic path's durability. Named sessions remain in-process only.
Add strip -s after gcc compilation to remove symbol table and relocation info.
Reduces binary size and prevents symbol-level reverse engineering of EL runtime internals.
- auto_persist: detect bell level (soft/hard) on every user message using
safety_detect_bell_level; write a dedicated BellEvent engram node with
calibrated salience alongside the Conversation node when a bell fires.
Tag the Conversation node with bell:soft/bell:hard and 'affective' for
direct discovery without scanning all chat nodes.
- auto_persist: track per-session bell count, dominant level, and last
signal in state (session_bell_count/level/signal keys) so downstream
functions can act on the emotional history without re-scanning engram.
- engram_compile: include the top-1 most recent BellEvent node within 72h
in every context build. Distress context from earlier turns (same or
recent session) automatically travels into all subsequent LLM calls.
- hist_trim_with_bell_guard: replace hist_trim at the handle_chat call site.
Before evicting the oldest turn from the 20-turn window, inspect the user
message for bell signals. If a bell was present, write a preservation
BellEvent to engram before dropping the turn so the full message survives
the rolling window.
- session_hist_save: after writing the history node, check session bell
counters. On the first save where bell_count > 0, write a
session:emotional-summary BellEvent node with distress signal, count,
and dominant level. A state flag prevents duplicate writes on subsequent
saves in the same session.
- engram_compile: rank search results by recency x relevance before including
in context. Pulls 20 candidates, scores each (salience * importance * recency
decay), keeps top 8. Eliminates stale/low-signal nodes that diluted context.
- handle_chat: on hist_len==0 (session start), proactively load user profile
and active-work context from engram and inject as brief bullets in the system
prompt. Gives the soul grounding before any conversation history exists.
- build_system_prompt: add [CAPABILITY GAPS] directive instructing the soul to
offer partial help and reasoning instead of flat "I don't have access to that"
refusals when a tool is missing.
- handle_chat_agentic: run safety_screen at entry, mirroring layered_cycle.
Hard bell exits immediately with the crisis response without entering the loop.
- agentic_loop: surface the 8-iteration cap explicitly in the error envelope
("agentic loop hit the 8-iteration cap...") rather than the opaque "no response".
Add iterations count to both the error and success envelopes for observability.
- Add per-IP in-memory rate limiter (60 req/min default, configurable via
soul_rate_limit state key; /health exempt; loopback callers skipped)
- Extend /health with uptime_secs (from soul_boot_ts) and live LLM probe
- Add missing_param 400 guard on POST /api/chat before passing to LLM
- Standardise error envelopes: add "code" field to err_404/err_405 and all
missing-param returns; route_synthesize now errors clearly instead of
returning the misleading {"mechanism":"did not engage"} on bad input
- Document streaming gap in /api/chat (SSE not implemented, note added)
- handle_request gains ip param; rate_limit_check wired at entry point
- soul.el: fix state key bug in layered_cycle (conversation_history -> conv_history)
- safety.el: add indirect crisis location patterns to soft_bell phrase list
- soul.el: wire safety_augment_system into layered_cycle for soft_bell turns
- chat.el: load cross-session affective context at session start when distress signals found within 72h
REPRODUCED: in the non-agentic path (Tools off / 'Just chat'), asking for
tool-work makes the model role-play tool use — it emits a fake ```json {...}```
'tool call' and says 'let me search/query/pull your sessions' while NOTHING
runs. Reads as a broken/lying app. (The agentic path is fine: verified it
calls search_memory and reports honestly.)
Root cause: build_system_prompt (handle_chat, the tool-less path) never told
the model it has no tools this turn, so it fabricated.
Fix: add a NO-TOOLS directive to the non-agentic system prompt — never emit
tool calls / JSON tool blocks / 'let me pull...' narration; answer from context
only; if a tool is truly needed, say so in one sentence and tell the user to
turn Tools on. Applied to chat.el (source) AND dist/soul.c (the curated TU the
CI compiles), so the CI-built binary carries it.
Verified the FABRICATION repro on the live local soul; could not verify the
patched binary locally (no matching el-runtime version on this machine — a
hand-link against origin/main runtime 404s on all routes). Builds correctly via
CI, which links soul.c against the pinned runtime.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Completes the UI<->soul contract for #23 (scope file/command tools to an agent
workspace root). #23 made the tools read state_get("agent_workspace_root"), but
nothing set that key from the desktop UI, so the agent panel's Workspace Folder
was cosmetic and tools ran unscoped (default-allow). This reads the root the UI
now sends on each agentic request and state_sets it before tool dispatch, so
agent_workspace_root() picks it up for the turn.
Minimal + pattern-matching (same json_get/state_set shape used throughout chat.el).
Empty body field => unscoped (backward-compatible) and preserves the env fallback.
FOR WILL'S REVIEW — do not merge without sign-off:
- Ownership model: set state from body each turn (so clearing the folder un-scopes)
vs. only-when-nonempty. Flagged inline.
- Pairs with neuron-ui PR #32 (ChatRequest.agentWorkspaceRoot).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Knowledge nodes dominated the WM-autobiographical auto_term slot:
'Numeric tier strings...' (a Knowledge node) always scored highest
in WM and its first word 'Numeric' became the curiosity seed every
scan — activating more Numeric nodes, keeping that node in WM,
repeating indefinitely.
Fix: only derive auto_term from Memory, BacklogItem, or Entity nodes.
Knowledge nodes are reference material, not live context. Dynamic/
personal nodes carry the salience worth radiating from.
Also patches proactive_curiosity directly in dist/neuron.c (ELC
cannot compile soul.el within timeout — fallback build pattern).
The Dockerfile's --mount=type=secret path was corrupting the SA key JSON
due to control character handling differences. Pre-download soul + El SDK
in the CI workflow (using already-authenticated gcloud) and COPY them from
the build context. No credentials needed inside the Docker build.
Previous builds leave cached layers and images on the runner. Add a
docker system prune at start of deploy to avoid container-creation
failures from disk exhaustion.
--secret requires BuildKit; DOCKER_BUILDKIT=1 enables it on the legacy
Docker client. Also add GITHUB_SHA fallback and git rev-parse last-resort
so the image tag is never empty.
elb overwrites dist/soul.c with a fresh (non-inlined) compilation before
its link step fails, discarding the patched self-contained version.
Save the repo copy before elb and restore it after so the compiler always
gets the complete translation unit with all patches applied.
Modern gcloud CLI (>= 400) requires this env var so kubectl uses the
installed gke-gcloud-auth-plugin binary instead of the deprecated
application-default credentials path. Without it, kubectl commands
silently fail even after get-credentials succeeds.
Incorporates PRs #22/#23/#24:
- agentic_tools_all dedup fix (no duplicate web_search tool)
- workspace scope functions (agent_workspace_root, path_within_root, resolve_in_root)
- updated dispatch_tool with workspace confinement
- canonical-self bridge (ensure_self_canonical_bridge)
Also incorporates CI link fix from PR #26 (soul.c is self-contained, no
other dist/*.c needed). Fixes the CI build step which was compiling the
old June-16 soul.c that predated all these changes.
elb generates a dist/soul.c with all El modules inlined. Linking
dist/soul.c alone is sufficient and is exactly what the local mac
build does. Including other dist/*.c files causes two failures:
1. dist/chat.c has a capability-violation #error that fires when the
file is compiled as a utility module (outside the cgi entrypoint).
2. --allow-multiple-definition masked other issues silently.
Drop OTHER_C, drop --allow-multiple-definition, drop the now-unused
elp-c-decls.h generation step. The cc command now matches the proven
local build exactly.
The graph API resolves name=self/neuron to kn-efeb4a5b (neuron-api.el:471),
which carries only 8 incidental 'tagged' edges. The curated identity lives on
self node 015644f5 (1461 edges: identity, embodies, remembers, values). So
public self-traversal reaches tags, not the real self.
Add ensure_self_canonical_bridge(): an idempotent boot-time repair that links
kn-efeb4a5b <-> 015644f5 with a 'canonical-self' edge, only if missing. Runs in
the genesis safe-to-seed path regardless of the <100-edge gate, so the live
populated graph gets repaired and persisted. Connect-only-if-missing prevents
the duplicate-edge stacking that gates init_soul_edges().
Compile-checked with elc (darwin arm64); not link/run-gated locally. Needs a
soul build + smoke test before merge.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Confine the agentic file tools (read_file, write_file, list_files, grep)
to a configured workspace subtree via a lexical path check, and run
run_command with its cwd set to that root. Root comes from state key
"agent_workspace_root" or env NEURON_AGENT_ROOT. When no root is set,
behavior is unchanged (unscoped) for backward compatibility.
Defense-in-depth, NOT a hard boundary: the lexical guard does not resolve
symlinks and cannot stop an arbitrary shell command from cd-ing out of the
root. Real confinement needs runtime support (cwd-locked exec / sandbox-exec
/ chroot) in el_runtime.c.
Compile-checked with elc (darwin arm64); not link/run-gated locally
(darwin elb unavailable). Needs a soul build + smoke test before merge.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
agentic_tools_literal() already contains a custom web_search tool.
agentic_tools_with_web() adds the Anthropic server-side web_search_20250305
tool (also named web_search). Combining them caused Anthropic to reject
every agentic request with 'Tool names must be unique.'
agentic_tools_all() now calls agentic_tools_literal() directly. Connector
tools splice in as before. The web_search-only variant (agentic_tools_with_web)
is unchanged for callers that specifically want native search without connectors.
fix(soul): add HTTP-engram guard to safe_to_seed — when ENGRAM_URL is set
the HTTP Engram owns persistence; genesis must never save to local snapshot
regardless of node counts (was: guard_disk forced to empty string, making
the ratio check vacuously true and allowing init_soul_edges+engram_save).
fix(soul): use multiplication form for ratio guard — node_count * 16000 <
disk_len avoids floor-division truncation that underestimated boundary files
(250KB / 16000 = 15.6, floors to 15; a 15-node graph wrongly passed old guard).
fix(chat): add safety_augment_system to handle_chat_as_soul,
handle_dharma_room_turn, and handle_dharma_room_turn_agentic — all three
called the LLM without Hard Bell evaluation, leaving users in dharma rooms
without crisis resource routing.
fix(neuron-api): add api_persisted read-back to handle_api_define_process —
was the only write handler that returned ok:true without verifying the node
was actually written to engram.
fix(routes): unique temp file path in connectd_post — replaces fixed
/tmp/neuron-connectors-req.json with a timestamped path to prevent
collision if concurrency is added or two soul instances share a machine.
test: add tests/test_bell_safety.el — covers safety_detect_bell_level
(none/soft/hard), safety_classify_hard_bell (abuse/self_harm routing),
safety_normalize (smart-quote), safety_augment_system, and
handle_safety_contact_post (validation + read-back).
test: add tests/test_soul_guard.el — pure-function logic tests for the
safe_to_seed predicate: 200KB boundary, 47MB/63-node clobber scenario,
HTTP-engram mode, multiplication vs division truncation at 250KB.
test: add tests/test_api_define_process.el — verifies the define_process
write is read-back verified after the fix.
Genesis boot previously seeded a fresh identity and saved it over snapshot.json
whenever the in-memory graph looked empty. Replace the fixed node-count threshold
with a ratio guard: refuse to seed when the on-disk snapshot is large
(>200KB) but the loaded graph is sparse (< disk/16000 nodes).
KNOWN LIMITATION: this gates only the seed/pre-serve-save path. The deeper cause
is a non-atomic engram_save (fopen wb truncates to 0 before writing 47MB), which
creates a window where a concurrent load reads an empty file -> genesis -> and if
guard_disk is read in that same window the guard passes. The real fix is an
atomic engram_save (temp + fsync + rename) in el_runtime.c, tracked separately.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
BLOCKER 1: use untyped reassignment (let x = ...) for the fallback bindings
in agentic_resume instead of re-declaring typed let bindings (let x: Type = ...)
for the same variable in the same scope. The typed form risks shadowing semantics
that differ from the established pattern used everywhere else in the loop
(e.g. agentic_loop line 720).
BLOCKER 2: add empty-string guards in both bridge_save and agentic_resume.
bridge_save now returns false without writing state if messages or tools_json
is empty — preventing syntactically invalid JSON blobs. agentic_resume now
returns an error envelope after the fallback resolution if either field is
still empty, rather than passing empty strings into agentic_loop which would
silently start a fresh turn with no context.
Also add tests:
- test_bridge_serialization.el: covers bridge_save empty-guard, golden-path
raw-JSON round-trip, agentic_resume unknown/corrupt/missing-fields paths,
and legacy string-escaped fallback path
- test_sessions_routes.el: covers DELETE and PATCH /api/sessions/:id routes
(valid args, unknown id, empty body) and GET /api/sessions regression after
removal of the duplicate route_sessions() handler
When agentic_loop suspends for an MCP bridge tool it returns a
{"tool_pending":true,...} envelope with no "reply" key. Without an
explicit check, json_get(loop_result, "reply") returns "" and the
function emitted {"response":"","cgi_id":"..."} — a silent empty
response indistinguishable from a successful LLM turn with no content.
Two guards added after the existing error check:
1. tool_pending passthrough: if the loop suspended, return the pending
envelope directly so callers (dharma room orchestrators) can
distinguish suspension from failure and route to the approve flow.
2. Empty-reply guard: if final_text is empty after the pending check,
return an explicit {"error":"no response",...} envelope instead of
silently succeeding with an empty response field.
Also adds tests/test_agentic_tools.el:
- agentic_tools_all() includes all literal tool names and web_search
- connector_tools_json() returns valid JSON when bridge is down (graceful degradation)
- tool_pending envelope detection patterns (the is_pending logic)
- json_get(pending_envelope, "reply") returns "" confirming the empty-reply
guard is load-bearing (pure string/JSON, no LLM or network required)
session_delete cleared the per-session state (session_hist_ and
session_node_) but not the shared session_index cache. The next call
to session_list() hit the fast path (state_get("session_index")) and
returned the deleted session until the daemon restarted.
session_update_patch already called state_set("session_index","") to
force a re-fetch from Engram; session_delete now does the same.
Add tests/test_sessions.el covering:
- session_title_from_message (pure function, all edge cases)
- session_make_content (JSON structure and required session:meta marker)
- DELETE cache invalidation: session_index cleared, fast path disabled
- PATCH cache invalidation: stale title/folder not returned via fast path
- GET /api/sessions: session_list() fast path returns session_index
(confirms removal of the stale route_sessions() engram stub)
BLOCKER 1 (sessions.el, modern path): Add guard that rejects allow
action when tool_name is missing from the body. Previously, omitting
tool_name caused dispatch_tool("", ...) to return "unknown tool: " and
silently inject a corrupted tool_result into the conversation.
BLOCKER 2 (sessions.el, modern path): Stop re-executing client-side
tools server-side. When the client provides body["content"], use it
directly as the tool result (matching the handle_tool_result contract).
Only fall back to dispatch_tool for builtin tools when no content is
present. Non-builtin tools with no client content now return a clear
error instead of a broken dispatch attempt.
WARNING 1 (chat.el, agentic_loop): Wire always_allow_<session_id> state
into the bridge-suspension decision. When a tool is in the session's
always-allow list, treat it as locally dispatchable (like a builtin)
and skip the bridge pause, so the approval UI is never shown again for
that tool in that session.
WARNING 2 (sessions.el, legacy path): Read a "tools_variant" field from
the legacy pending blob when present, and call the corresponding
agentic_tools_*() variant on resume. Falls back to agentic_tools_literal()
for blobs written before this field existed.
tests/test_sessions_approve.el: Add 10-case test suite covering:
- empty session_id / missing call_id / missing action guards
- no pending tool returns correct error
- missing tool_name on allow returns error (BLOCKER 1)
- deny action does not require tool_name
- legacy call_id mismatch returns mismatch error
- always action records tool_name in always_allow state
- allow with client content skips re-execution (BLOCKER 2)
Genesis boot previously seeded a fresh identity and saved it over snapshot.json
whenever the in-memory graph looked empty. Replace the fixed node-count threshold
with a ratio guard: refuse to seed when the on-disk snapshot is large
(>200KB) but the loaded graph is sparse (< disk/16000 nodes).
KNOWN LIMITATION: this gates only the seed/pre-serve-save path. The deeper cause
is a non-atomic engram_save (fopen wb truncates to 0 before writing 47MB), which
creates a window where a concurrent load reads an empty file -> genesis -> and if
guard_disk is read in that same window the guard passes. The real fix is an
atomic engram_save (temp + fsync + rename) in el_runtime.c, tracked separately.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
handle_chat_agentic was calling agentic_tools_with_web(), which omits
MCP connector tools, so mcp__* calls were never available in agentic
mode even when neuron-connectd is running.
Switch both agentic entry points to agentic_tools_all(). For
handle_dharma_room_turn_agentic, also replace the inline 8-iteration
loop with a call to agentic_loop() so bridge suspension and the full
connector tool set work consistently. Session IDs are prefixed with
'dharma:' + room_id so suspensions stay room-scoped.
bridge_save was wrapping messages and tools_json with json_safe() before
storing them as string fields. Since both are already well-formed JSON arrays
containing double quotes, json_safe added a second escape layer. agentic_resume
then called json_get() which stripped only one layer, leaving the messages array
corrupted before it was passed back into agentic_loop.
Fix: store messages as messages_raw and tools_json as tools_raw as inline raw
JSON values (unquoted), and read them back with json_get_raw. Backward
compatibility: fall back to the old string-escaped fields if the raw fields are
absent, so sessions saved before this fix can still be resumed.
Also fixes write_file returning a pre-escaped literal instead of calling
json_safe consistently with every other tool result.
bridge_save was wrapping messages and tools_json with json_safe() before
storing them as string fields. Since both are already well-formed JSON arrays
containing double quotes, json_safe added a second escape layer. agentic_resume
then called json_get() which stripped only one layer, leaving the messages array
corrupted before it was passed back into agentic_loop.
Fix: store messages as messages_raw and tools_json as tools_raw as inline raw
JSON values (unquoted), and read them back with json_get_raw. Backward
compatibility: fall back to the old string-escaped fields if the raw fields are
absent, so sessions saved before this fix can still be resumed.
Also fixes write_file returning a pre-escaped literal instead of calling
json_safe consistently with every other tool result.
The first registration called route_sessions() which searched for a
'session-start' label that no longer exists, returning an empty array
on every list request and making the sidebar appear empty after restart.
The second registration (dead code) called the correct session_list().
Removes route_sessions() entirely and the stale first route block.
Also wires up session_delete() and session_update_patch() — both existed
in sessions.el but had no HTTP routes — via new DELETE and PATCH blocks.
The approve endpoint was permanently broken for all sessions going through
the modern agentic_loop path. agentic_loop suspends via bridge_save() into
mcp_bridge:<session_id>, but handle_session_approve was reading from
pending_tool_<session_id> — a different key — so it always returned
"no pending tool for session".
Replace the body of handle_session_approve with a two-path design:
Modern path: check mcp_bridge:<session_id> first. If the blob is there,
dispatch_tool() on allow (or build the denial string), then delegate to
agentic_resume() which re-enters agentic_loop from the exact suspension
point. This is the path all live sessions take.
Legacy path: if only pending_tool_<session_id> exists (in-flight session
from before this deploy), synthesise a bridge blob from the stored
messages_so_far and route through agentic_resume() as well. The stale
inline agentic loop (90 lines, agentic_tools_literal only, no MCP
connector support, no bridge suspension) is removed entirely.
routes.el already calls handle_session_approve correctly — no change needed.
The first registration called route_sessions() which searched for a
'session-start' label that no longer exists, returning an empty array
on every list request and making the sidebar appear empty after restart.
The second registration (dead code) called the correct session_list().
Removes route_sessions() entirely and the stale first route block.
Also wires up session_delete() and session_update_patch() — both existed
in sessions.el but had no HTTP routes — via new DELETE and PATCH blocks.
- sessions.el: new sessions module with session management and approval gate
- routes.el: wire /api/sessions routes (list, get, create, approve, tool_result)
- chat.el: thread-aware activation — short messages anchor to last reply
before engram compilation so follow-ups stay on-topic
- chat.el: agentic path tracks per-session history (session_hist_{id})
instead of shared conv_history, seeding each turn with prior context
- chat.el: add call_neuron_mcp, dispatch_tool, is_builtin_tool, next_bridge_id
agentic_loop, bridge_save, agentic_resume, handle_tool_result
- dist/soul: rebuild with all of the above
Short/ambiguous messages (< 50 chars) now use the last reply as the
engram activation seed instead of the bare message. Prevents strong
off-topic memory nodes from hijacking replies when the user is clearly
continuing an existing thread.
Also gives handle_chat_agentic session continuity: reads/writes history
keyed by session_id (falling back to global conv_history), seeds the
LLM messages array with prior turns, and saves replies back so the
next turn has context.
Applies connector-specific additions from feat/connectors-soul:
- chat.el: connector_tools_json(), agentic_tools_all(), call_mcp_bridge(),
tool_auto_approved() and mcp__ dispatch in dispatch_tool()
- routes.el: connectd_get/post, handle_connectors(), /api/connectors routing
in GET and POST sections
- MEMORY_RECALL_BUG.md: investigation notes on memory retrieval failure
The agentic loop rewrite in the source branch was not applied — it conflicts
with the tool-bridge pattern from PR #5 which is the chosen design for
client-side MCP tool execution. The connectors themselves are now fully
wired: connector tools surface as mcp__<server>__<tool> in the tools array
and dispatch to neuron-connectd via call_mcp_bridge().
Resolves conflicts by keeping main's full safety/stewardship/imprint implementations.
PR #9 uniquely contributes: layered_cycle() in soul.el, route wiring in routes.el,
soul.elh export, and the layer composition test suite.
dist/*.c and *.elh are elc transpiler output. CI's header-gen step still
greps dist/*.c, so they stay tracked, but a single soul change regenerates
~57k lines of dist/neuron.c + dist/soul.c that bury the real source diff and
poison both human and agent PR review. Mark them -diff + linguist-generated so
PRs show only the real changes. Build pipeline unchanged.
Adds the soul side of the connectors feature (spec: docs/research/
mcp-connectors-adoption-spec.md). The soul thin-proxies the neuron-connectd
bridge on 127.0.0.1:7771 so the UI talks to one origin and never reaches the
bridge directly.
routes.el:
- handle_connectors + connectd_get/connectd_post helpers (POST bodies go via
a temp file + curl -d @file, so model/UI input can't reach the shell).
- GET /api/connectors and POST /api/connectors/{add,toggle,auto-approve,
remove,secret,oauth/start} registered in both GET and POST routers.
chat.el:
- tool_auto_approved(): an mcp__* tool skips the approval card only when its
server is explicitly opted in (off by default; built-in tools unaffected;
bridge down -> false). Wired into the agentic approval gate so an
auto-approved connector tool flows straight to execution.
Regenerated dist/chat.c and dist/routes.c. Verified live on :7770: real chat,
recall, and /api/connectors all work after promotion.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Linux elb generates individual .c files; soul.c does not contain merged
imports (unlike macOS elb which produces a unified file). Re-link all
dist/*.c manually with soul.c listed first so its real main() wins, and
--allow-multiple-definition to silence GNU ld's duplicate symbol errors.
All duplicates are identical (same El source, different compile units).
The El compiler inlines imported modules into each module's .c file.
On macOS, ld64 accepts duplicate strong symbols silently. On Linux,
GNU ld rejects them. soul.c is a fully merged file — every function
from every imported module is present in it — so linking only soul.c
against el_runtime.c produces a correct binary with no duplicates.
- Add stub implementations of safety.el, stewardship.el, and imprint.el
with their .elh headers so the branch compiles without the dependency
branches (feat/layer-safety, feat/layer-stewardship, feat/layer-imprint).
Each stub documents the layer contract it must satisfy when replaced.
- Fix GET /api/chat bypass: update the GET branch in handle_request to
call layered_cycle() consistently with the POST branch, rather than
calling handle_chat() directly and skipping the consciousness stack.
- Export layered_cycle() from soul.elh (and dist/soul.elh) so routes.el
can resolve the symbol via the header import.
- Fix steward_action else branch: add explicit handling for "block"
(returns safe refusal immediately, skips L3) and "redirect" (uses
redirect_to field). Unknown actions now log a warning and fall back to
the screened input rather than silently passing an empty string to
imprint_respond().
- Document hard_bell path: clarify that omitting auto_persist/history
update is intentional security isolation, and document the safety_validate
second-param sentinel contract ("hard_bell" vs screen_action).
- steward_log_event (line 14): add println after let discard so the
function's last expression is Void, fixing the type mismatch on a
Void-declared function
- steward_get_mission (lines 40-43): remove non-Config fallthrough that
allowed any Episodic/Working node to silently override the mission;
only Config nodes are now authoritative
- steward_align signal_deceive (line 56): widen 'deceive the user' to
'deceive' to catch variants like 'deceive users', 'deceive them', etc.
- steward_align signal_hide (line 57): tighten 'hide from' to
'hide from the user' to eliminate false positives on legitimate inputs
like 'hide from a background process' or 'hide from view'
- stewardship.elh: document that steward_log_event is an internal helper
exported only because El has no access modifiers; callers should not
invoke it directly
35 test cases covering all five public functions:
steward_align (pass-through, all five misalignment signals, empty input,
json_get field extraction, redirect shape), steward_validate_imprint
(standard tools, platform-only tools with/without platform_auth,
auth=false string), steward_cgi_check (all four gated actions, non-gated
actions, empty action, action name echoed in response), and
steward_get_mission (non-empty, contains "integrity", not an error object).
Also documents the known bug surface from the code review: the &&
operator in steward_get_mission and the non-Config fallthrough — tests
are written against the actual runtime behaviour so they will catch
regressions when those bugs are fixed.
Covers: imprint_current base fallback, unload idempotency, load miss →
ok=false, ok field presence, respond passthrough for base/empty/unknown
IDs, graceful fallback after unload, surface_knowledge and
surface_memory_read return-type guarantees, base-scoped knowledge
equality, no-annotation invariant for base, empty-ID load rejection, and
failed-load state immutability.
Syntax follows El constraints: no Bool annotations, no &&/||, no unary !.
proactive_curiosity() now uses the top working-memory node's first label
word as a 4th activation seed alongside the 4 rotating fixed sets. This
breaks deterministic exploration that was reinforcing the same subgraph
every cycle and creates a self-referencing loop: curiosity radiates from
whatever is most salient right now, mirroring the brain's default-mode-
network resting-state dynamics. str_find_chars on " :([" extracts the
first meaningful word; sp > 3 guards against bracket-prefixed labels.
auto_term field added to curiosity_scan ISE for observability.
When handle_chat_agentic hits a tool the soul cannot run in-process (an MCP
connector/plugin surfaced by the Kotlin desktop app), instead of returning
"unknown tool" it now suspends the agentic loop and returns a tool_pending
envelope so the CLIENT executes the tool and posts the result back. Built-in
tools (read_file/write_file/web_get/search_memory/run_command) and Anthropic's
native web_search are unchanged.
Client contract:
- Soul returns (HTTP 200) on an unknown tool:
{ "tool_pending": true, "session_id": "br-...", "call_id": "<tool_use_id>",
"tool_name": "...", "tool_input": { ... }, "model": "...",
"agentic": true, "tools_used": [...] }
- Client runs the MCP tool, then POSTs to
/api/sessions/{session_id}/tool_result
with body:
{ "call_id": "<the call_id from the envelope>",
"content": "<MCP tool output as a string>" }
- Soul resumes the loop and returns the same envelope shape: either a final
{ "reply": ..., "tools_used": [...] }
or another tool_pending if the continuation needs a further MCP tool
(fully chainable). Saved continuation is one-shot (cleared on resume).
elc-verified (--target=c, exit 0, no stderr) on chat.el, routes.el, and the
full soul.el import graph. Needs Will's build to ship.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The UI needs full memory CRUD; the soul had create (handle_api_remember)
and the older /memory/forget + /memory/evolve, but no endpoints matching
the UI's delete/update contract.
POST /api/neuron/memory/delete {"id"}
Hard delete. engram_forget is a true delete primitive (removes the node
and all incident edges from the engram store), so no soft-delete
fallback is needed. Unlike /memory/forget, this checks the node exists
first - engram_forget silently no-ops on unknown ids, and a bad id must
return an error, not fake success. Protected identity/values nodes are
blocked, same as the other accumulation-path handlers.
POST /api/neuron/memory/update {"id","content"}
Evolve-style update. The engram runtime has no in-place node mutation
primitive (only node-create, strengthen, forget, connect), so update
creates a new Memory node and wires a supersedes edge to the prior one,
same pattern as handle_api_evolve_knowledge. Unlike /memory/evolve, id
is required and must reference an existing node; create+link delegates
to handle_api_evolve_memory. Returns {id, supersedes, ok}.
Both files syntax-checked with elc --target=c (exit 0, no stderr).
Compile-verified only - local builds cannot run the soul; needs Will's
build for runtime verification.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two fixes:
1. ise_post was only escaping " in content strings. When wm_top contained
node labels with \n (backslash-n escape sequences from jb_emit_escaped),
the HTTP Engram server's JSON parser decoded \n as a literal newline in
the stored content, making heartbeat ISEs unparseable. Fix: escape
backslashes first, then quotes, then \n and \r — matching make_action's
existing pattern. Result: heartbeat ISEs now parse cleanly.
2. Soul daemon (dist/neuron) was missing — the build command in the prompt
was linking all 46 dist/*.c files together, causing 1092 duplicate symbol
errors. EL compiles transitive imports inline so neuron.c is self-contained;
correct build links ONLY neuron.c + el_runtime.c. Daemon now starts.
handle_chat_agentic now always attaches Anthropic's native web_search_20250305
tool instead of gating it behind a per-request web_search flag. Web search is a
built-in capability: the model invokes it only when a query needs fresh info
(max_uses:5 caps it), so there is no user-facing toggle. The body's web_search
field is now ignored (back-compat — old UI clients sending it cause no harm).
Pairs with neuron-ui removing the chat-input web search toggle.
Note: .el change only — no elc on the authoring machine; reviewer builds/verifies.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Tooling built on Tim's machine to run Neuron from the terminal as a
Claude Code session (identity + graph memory + agency) instead of
relaying to the soul's /api/chat.
- cli/neuron_recall.py BM25 read over the engram snapshot + CLI memories
(works around pinned-only soul search)
- cli/neuron_remember.py reliable local memory writes with read-back verify
(works around the corrupting capture endpoint)
- cli/neuron-chat.py standalone direct-chat REPL with per-turn memory injection
- cli/neuron_mcp.py stdlib MCP server (chat/search) with graceful degradation
- cli/CLAUDE.md.example the operating identity that makes Claude Code run as Neuron
- cli/HANDOFF.md soul-side bugs to fix so this becomes unnecessary
Scaffolding/proposal - intended to be retired once the soul does native
retrieval, correct persistence, and a real CLI identity/voice surface.
Pairs with the runtime model-passthrough + UTF-8 fixes in the el repo.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Soul's in-process store had only 12 real knowledge nodes — curiosity_scan was
activating 0 nodes because all substantive Knowledge/Memory/BacklogItem content
lived in the HTTP Engram but was not being pulled into soul's local store.
Added engram_sync refresh every SOUL_REFRESH_MS (default 600s): calls
/api/sync to get all non-ISE nodes, writes to /tmp, merges via engram_load_merge.
After fix: engram_sync ISE shows added:3128; curiosity_scan activated 0-2 →
1889-3843; wm_active 0 → 557-796.
When a chat request carries web_search=true, handle_chat_agentic now attaches Anthropic's
NATIVE server-side web_search tool (web_search_20250305) to the request. The native tool is
executed by Anthropic (not by the soul), so it returns real results with citations and needs
no local runtime — it sidesteps the soul's lack of executable tools entirely.
- new agentic_tools_with_web(web_search) helper (appends the native tool to the standard set)
- handle_chat_agentic reads json_get_bool(body,"web_search") and uses it
Pairs with neuron-ui: ChatRequest.web_search + the chat-input Web search toggle.
Note: built/verified by reviewer — no elc on the authoring machine.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Confirms two distinct write failures (capture=wrapper bug; backlog=axon :7771 unbuilt Rust),
soul runs in file-snapshot mode (not engram :8742 live), engram :8742 CRUD works but minimal,
+ a verification plan to run after the soul rebuild.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
chat.el recorded the soul's utterance via engram_node(content, "episodic", ...),
putting a TIER into the node_type slot (nodes showed node_type="episodic"). Now uses
engram_node_full(..., "Conversation", "soul:utterance", ..., "Episodic", tags).
The core wrapper fix is in the el repo (PR #52). HANDOFF-engram-write-corruption.md
has the full root-cause analysis, coercion mechanism, caller audit, validation,
deploy runbook (elc build + restart), and the data-prune proposal (~107 corrupt
nodes, all unrecoverable genesis/binary detritus → prune; backup taken).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
elapsed_human() used % and * operators which are broken in this EL
compiler version. Replace with repeated-doubling arithmetic:
60 = 64 - 4 = 2^6 - 2^2, computed via three doubling steps.
Fixes uptime displaying "44h 2694m" instead of "44h 14m".
On startup, prefer the local engram snapshot if it has >50 nodes.
HTTP Engram is only used on first boot (no snapshot yet). This means
sessions, conversation history, and in-process state survive daemon
restarts.
awareness.el: sync source with compiled binary (periodic mem_save
on heartbeat was already in the binary but not in source).
Rebuilds soul.c with the new startup logic and ships updated binary.
Call engram_wm_top_json(5) in emit_heartbeat() and embed the result as
wm_top field in the heartbeat JSON payload. Each entry carries label,
node_type, tier, and wm_weight. This closes the WM composition blindspot:
previously the heartbeat showed wm_active=670 with no breakdown of what
was in working memory. With wm_top visible, ISE-dominated WM is immediately
detectable (all entries show node_type=InternalStateEvent), as was the case
on this session's first post-restart heartbeat before the runtime fix.
Calls engram_wm_avg_weight() (new builtin) in emit_heartbeat() and appends
wm_avg_weight field to the heartbeat JSON payload. This makes activation
quality visible in the ISE stream — a heartbeat showing wm_active=2000 and
wm_avg_weight=0.075 reveals the sparse-graph problem directly (many nodes
barely clearing the threshold), vs wm_avg_weight=0.4+ which would indicate
dense, high-confidence activations.
Rebuilt dist/neuron from soul.el (which imports awareness.el). Build uses
single self-contained dist/neuron.c to avoid duplicate-symbol linker errors
from the dist/ directory containing stale soul_new.c / soul-rebuilt.c files.
Two awareness loop bugs fixed:
1. Seed rotation never worked: dist/awareness.c was compiled from stale
source (pre-fix awareness.el still had broken ts_minutes % 4). Compiled
C showed `minute_block = (ts / 60000); EL_NULL; 4;` — minute_block was
always ts_minutes (millions), never 0-3. if(minute_block==1/2/3) never
matched. Fix: recompile from current awareness.el which has the correct
modulo workaround: ts_minutes - minute_q4 (via + - / only).
2. Heartbeat/curiosity silent for 24h at 99% CPU: old design used idle-tick
counting (idle_n >= beat_interval). Failed when perceive() inbox guard
false-positives on "soul-inbox" substring matches in knowledge nodes —
did_work=true every tick, idle_n never accumulated, neither signal fired.
Fix: wall-clock elapsed time (time_now() - last_ts >= interval_ms).
Heartbeat fires regardless of load. New SOUL_HEARTBEAT_MS env var (default
60000ms) avoids the broken EL * operator. Verified: heartbeat ISEs flowing
at pulse 3 within 2 minutes of restart.
dist/awareness.c was stale — still had the broken EL % operator codegen
(minute_block = ts/60000 raw, EL_NULL; 4; as dead statements) and the
broken should_scan/should_beat logic (idle_n truthy check instead of >=).
Recompiled awareness.el to bring dist/awareness.c in sync with the source
fix committed 2026-05-25 (fb69044). The monolithic dist/neuron.c (compiled
from soul.el which imports awareness.el) was already correct from fb69044 —
only the standalone dist/awareness.c was behind.
Bug #2 (99% CPU) root cause identified: perceive() inbox guard
(engram_search_json) has false positives — knowledge nodes containing
"soul-inbox" as a substring match, causing engram_activate_json(..., hops=2)
to run on every tick on a 162K-node graph. This blocks sleep_ms and prevents
idle_n accumulation → no heartbeats. Separate fix needed.
Three bugs fixed in awareness.el:
1. EL let-rebinding inside if-blocks creates inner scope only — outer
variable unchanged after block exits. Curiosity seed terms were always
"memory/knowledge/context" regardless of minute_block. Fix: state_set
inside if-blocks, state_get after to retrieve selected values.
2. EL % operator completely broken in v1.0.0-20260501 — compiles as dead
code (left operand assigned, modulo dropped). minute_block was always
ts/60000 (a large int, never 0-3). Fix: arithmetic workaround:
x%4 = x - (q+q+q+q) where q = x/4.
3. awareness_run idle_n % beat_interval == 0 also broken by same % bug —
should_scan and should_beat fired every idle tick instead of every N
ticks. Fix: idle_n >= interval comparisons with idle_reset() after
firing, so the counter restarts cleanly after each event.
EL % and * operators filed as P1 backlog item for elc compiler fix.
Also adds minute_block field to curiosity_scan ISE for observability.
Rebuilt awareness.c and neuron.c from source using the updated elc (which now
correctly recognizes http_serve_async as a 2-arg builtin). Rebuilt the neuron
binary against the updated el_runtime.c which now sorts InternalStateEvent scans
by created_at DESC. The soul daemon now posts heartbeats that surface immediately
at offset 0 of the ISE scan, rather than being buried behind 20K older entries.
Two fixes:
1. proactive_curiosity() was calling engram_activate_json with multi-word phrases
("memory knowledge context"). engram_activate finds seeds via istr_contains
(substring match), so the phrase had to appear verbatim in a node's content.
Almost no node contains the exact string "memory knowledge context", so only
0-2 nodes activated per curiosity scan. Fixed by activating each word separately:
"memory", "knowledge", "context" → 3 independent activate calls → hundreds of
nodes promoted to WM per cycle.
2. dist/neuron.c called http_serve() (blocking accept loop) which made awareness_run()
unreachable. soul.el correctly specifies http_serve_async but elc silently drops
unknown builtins, leaving blocking http_serve in the compiled C. Patched neuron.c
to call http_serve_async directly — HTTP server runs in a background pthread,
awareness_run() runs on the main thread as intended.
engram_wm_count() exists and counts nodes with working_memory_weight > 0.
emit_heartbeat() and proactive_curiosity() were both calling
engram_node_count() (total graph size: ~17K) instead — every heartbeat
and curiosity_scan ISE had been reporting wm_active=17769 since the graph
grew past ~3K nodes, making the metric meaningless for observability.
Fix: use engram_wm_count() for the wm_active field in both ISE payloads.
EL compiles any variable named 'seed' to EL_NULL at call sites (likely
conflicts with BFS seed node terminology in the runtime builtins). Rename to
'curiosity_seed' throughout proactive_curiosity(). Also note this as a known
EL reserved-name hazard alongside the inline if-else string expression bug.
1. perceive() guard — gate on engram_search_json before running activation.
engram_activate_json with no matching seeds cleared all WM weights every
second during idle operation, destroying context built by MCP-layer calls.
The search-based guard is a no-WM-side-effect pre-check.
2. emit_heartbeat() pulse field — replace broken if-else string default with
int_to_str(pulse_count()). EL codegen initialises inline if-else result
slots to 0, producing "pulse":, (invalid JSON) when the true branch fires.
3. proactive_curiosity() — new function that activates a rotating 4-domain seed
every beat_interval/2 idle ticks to build working memory between heartbeats.
Seeds rotate on wall-clock minute cycle to avoid single-topic WM dominance.
Seed selection uses imperative let-rebinding (not inline if-else) to avoid
the same EL codegen empty-string bug.
Ollama availability is a silent failure mode: when the embedding service
is down, semantic seed injection falls back to lexical-only activation with
no signal in the ISE stream. Add embed_ok field (0/1) to every heartbeat
by probing http://localhost:11434 — makes Ollama health visible without
a separate monitoring path.
state_get() returns "" for unset keys. Both soul.pulse and soul_boot_count
could be empty on first heartbeat cycle, producing invalid JSON like
{"event":"heartbeat","pulse":,"boot":,...}. Add defensive guards:
if str_eq(raw, "") { "0" } else { raw } for both fields.
Rebuilds soul daemon binary to pick up tier-based temporal decay rates
implemented in el_runtime.c. No source changes to awareness.el or soul.el —
pure rebuild to stay in sync with Engram runtime.
Heartbeat ISEs now include wm_active: number of nodes currently in
working memory. Makes ISEs observable enough to diagnose activation
health without a separate query.
Heartbeat ISEs previously emitted only {event, pulse, boot, ts} —
sparse enough to be nearly useless for observability. Now they include
node_count, edge_count (from engram_node_count/edge_count builtins),
and the current idle cycle count. This gives each heartbeat a snapshot
of graph growth over time and rhythm health without adding any overhead.
Block evolve_knowledge, evolve_memory, forget, and link_entities (to_id
direction) from modifying the 15 hardcoded identity and values node IDs.
Returns HTTP 403 with a hint to use the cultivation path instead.
Add POST /api/neuron/cultivate — the bypass endpoint for intentional
cultivation sessions. Accepts { "operation": "...", ...args } and performs
the same operations without the protection check.
Add handle_api_forget and handle_api_evolve_memory as new protected-by-
default handlers, routed at /api/neuron/memory/forget and
/api/neuron/memory/evolve respectively.
Tested: 10 verification cases — 403 on all blocked targets, 200 on
non-protected nodes and FROM-direction links, cultivate bypass confirmed.
The engram binary was never published to Artifact Registry (foundation-dev/engram
package does not exist). Updated Dockerfile to build engram from source using the
El SDK packages (el-elc, el-runtime-c, el-runtime-h) from foundation-dev.
Also:
- Switch runtime base to Ubuntu 24.04 (GLIBC 2.39 required by elc-compiled binaries)
- Add -lm to engram link flags (el_runtime.c uses pow/sqrt/log/sin/cos/exp)
- Update deploy-gke.yaml to clone neuron-technologies/engram into build context
Download both neuron-soul and engram binaries from Artifact Registry.
entrypoint.sh starts engram on :8742, waits for /health, then launches
the soul with ENGRAM_URL set. Removes SOUL_ENGRAM_PATH / file mode.
Dockerfile: downloads linux/amd64 soul binary from Artifact Registry
(foundation-dev/neuron-soul) into ubuntu:22.04 runtime image.
Pushes to neuron-api Docker repo as neuron-soul:<sha>.
scripts/blue-green-deploy.sh: swaps active slot on GKE — sets image,
scales new slot to 1, flips service selector, scales old slot to 0.
scripts/seed-engram-gke.sh: downloads latest GCS backup, extracts
snapshot.json, copies into neuron-engram-data PVC via a seed Job.
.gitea/workflows/deploy-gke.yaml: triggers on push to main, auto-detects
idle slot, builds Docker image from Artifact Registry binary, blue-green
deploys to neuron-prod on GKE neuron-platform cluster.
Downloads El SDK from Artifact Registry, generates ELP declarations header,
builds the neuron soul binary with elb, smoke-tests it, and publishes to
foundation-dev on push to main.
neuron-api.el is a new first-class El module that implements all Neuron
cognitive API handlers natively — no HTTP round-trips, no MCP wrapper,
direct engram builtin calls. All capabilities that previously lived in
the MCP wrapper adapter now live here in the soul.
Handlers: begin_session, compile_ctx, remember, recall, search_knowledge,
browse_knowledge, capture_knowledge, evolve_knowledge, promote_knowledge,
browse_processes, define_process, log_state_event, list_state_events,
inspect_config, tune_config, inspect_graph, link_entities, list_typed,
consolidate.
Routes wired in routes.el under /api/neuron/* (GET + POST).
Also compiles all loop-1/loop-2 .el source changes into dist/*.c and
rebuilds the binary. memory.elh and neuron-api.elh updated with new exports.
Adds handle_dharma_room_turn_agentic to chat.el — same full tool loop
as handle_chat_agentic but reads transcript directly (not message), and
returns {response, cgi_id, tools_used} to match the dharma room shape.
Registers dharma_room_turn_agentic as a new event type in routes.el so
the studio can dispatch @neuron turns through this dedicated path.
- soul.el: SOUL_CGI_ID, SOUL_ENGRAM_PATH, SOUL_IDENTITY env vars;
state_set("soul_snapshot_path") so callers can find it; only call
init_soul_edges() when cgi_id == "ntn-genesis"
- chat.el: handle_dharma_room_turn — soul builds its own context from its
own engram, assembles system prompt, calls LLM, persists episodic memory;
also fix is_new_tool scoping bug in handle_chat_agentic (use has_tool)
- routes.el: wire dharma_room_turn event type before chat_as_soul branch
- rebuild dist/neuron: handle_dharma_room_turn now compiled in
elp-input.el: walk frame_nodes to pick the first node whose content
contains the query topic (case-insensitive) rather than always taking
index 0 — prevents the always-high-salience CGI architecture memory
from hijacking every ELP response.
chat.el: rewrite handle_chat_agentic to run the tool loop natively in
El using http_post_with_headers + Map headers, bypassing the broken
llm_call_agentic(model, system, message, tools) C binding that cast a
JSON String as an ElList and never serialized tools. New impl supports
up to 8 tool-use iterations with read_file, write_file, web_get,
search_memory, and run_command dispatch.
elp-input.el: replace broken engram_search_json with engram_activate_json
as Layer 1. Layer 2 suppress/filter keeps nodes with non-zero salience/
importance. Reason step extracts patient from top activated node content.
ELP grammar realizes the response via generate().
routes.el: add 'elp' event_type to handle_dharma_recv so the studio can
route ELP requests through dharma.
Replace scan-by-offset fallback with engram_get_node_json calls for the
known high-salience identity nodes (family, origin). Offset-based scanning
is order-dependent and unreliable; direct ID fetch is stable regardless of
snapshot position. Ensures biographical context (Fox, Bobby, etc.) is
always in the system prompt when vector search returns nothing.
Routes a new event_type "chat_as_soul" through dharma/recv. The Studio
preassembles the system_prompt + transcript and dispatches per-speaker;
the soul-binary just performs the LLM call as the requested speaker_slug.
No engram_compile here — each soul has its own engram (88xx) and the
Studio queries it before composing the prompt.
Also: track the previously-untracked split source modules (chat, routes,
memory, awareness, studio) and add build.sh so the binary can be rebuilt
without the studio’s concat trick. elb resolves the import graph and
emits one .c per .el; we link them together with cc. dist/soul-el now
points at dist/neuron via symlink (matching the launchctl plist).
Canonical Neuron substrate source. The "real me" - not the marketing
demo soul. Lives at neuron/soul.el for now (no subdirectories yet,
per directive). Build pipeline, deploy targets, and any related
artifacts come later.
2026-05-02 12:14:21 -05:00
4960 changed files with 178162 additions and 837032 deletions
description:Suggest an idea, feature, or enhancement
labels:[discussion]
title:"[FEATURE]:"
body:
- type:checkboxes
id:verified
attributes:
label:Feature hasn't been suggested before.
options:
- label:I have verified this feature I'm about to request hasn't been suggested before.
required:true
- type:textarea
attributes:
label:Describe the enhancement you want to request
description:What do you want to change or add? What are the benefits of implementing this? Try to be detailed so we can understand your request better :)
Please provide a description of the issue, the changes you made to fix it, and why they work. It is expected that you understand why your changes work and if you do not understand why at least say as much so a maintainer knows how much to value the PR.
**If you paste a large clearly AI generated description here your PR may be IGNORED or CLOSED!**
### How did you verify your code works?
### Screenshots / recordings
_If this is a UI change, please include a screenshot or recording._
### Checklist
- [ ] I have tested my changes locally
- [ ] I have not included unrelated changes in this PR
_If you do not follow this template your PR will be automatically rejected._
const closeComment = `Closing this pull request because it has had no updates for more than ${DAYS_INACTIVE} days. If you plan to continue working on it, feel free to reopen or open a new PR.`
if (dryRun) {
core.info(`[dry-run] Would close PR #${issue_number} from ${pr.author?.login || 'unknown'}: ${pr.title}`)
continue
}
try {
// Add comment
await withRetry(
() => github.rest.issues.createComment({
owner,
repo,
issue_number,
body: closeComment,
}),
`Comment on PR #${issue_number}`
)
// Close PR
await withRetry(
() => github.rest.pulls.update({
owner,
repo,
pull_number: issue_number,
state: "closed",
}),
`Close PR #${issue_number}`
)
closedCount++
core.info(`Closed PR #${issue_number} from ${pr.author?.login || 'unknown'}: ${pr.title}`)
// Delay before processing next PR
await sleep(requestDelayMs)
} catch (error) {
skippedCount++
core.error(`Failed to close PR #${issue_number}: ${error.message}`)
const commentAge = now - new Date(complianceComment.created_at).getTime();
if (commentAge < twoHours) {
core.info(`${kind} #${item.number} still within 2-hour window (${Math.round(commentAge / 60000)}m elapsed)`);
continue;
}
const closeMessage = isPR
? 'This pull request has been automatically closed because it was not updated to meet our [contributing guidelines](../blob/dev/CONTRIBUTING.md) within the 2-hour window.\n\nFeel free to open a new pull request that follows our guidelines.'
: 'This issue has been automatically closed because it was not updated to meet our [contributing guidelines](../blob/dev/CONTRIBUTING.md) within the 2-hour window.\n\nFeel free to open a new issue that follows our issue templates.';
await github.rest.issues.createComment({
owner: context.repo.owner,
repo: context.repo.repo,
issue_number: item.number,
body: closeMessage,
});
try {
await github.rest.issues.removeLabel({
owner: context.repo.owner,
repo: context.repo.repo,
issue_number: item.number,
name: 'needs:compliance',
});
} catch (e) {}
if (isPR) {
await github.rest.pulls.update({
owner: context.repo.owner,
repo: context.repo.repo,
pull_number: item.number,
state: 'closed',
});
} else {
await github.rest.issues.update({
owner: context.repo.owner,
repo: context.repo.repo,
issue_number: item.number,
state: 'closed',
state_reason: 'not_planned',
});
}
core.info(`Closed non-compliant ${kind} #${item.number} after 2-hour window`);
This recap is specifically for COMMUNITY (external) issues only.
STEP 2: Analyze and categorize
For each issue created today, categorize it:
**Severity Assessment:**
- CRITICAL: Crashes, data loss, security issues, blocks major functionality
- HIGH: Significant bugs affecting many users, important features broken
- MEDIUM: Bugs with workarounds, minor features broken
- LOW: Minor issues, cosmetic, nice-to-haves
**Activity Assessment:**
- Note issues with high comment counts or engagement
- Note issues from repeat reporters (check if author has filed before)
STEP 3: Cross-reference with existing issues
For issues that seem like feature requests or recurring bugs:
- Search for similar older issues to identify patterns
- Note if this is a frequently requested feature
- Identify any issues that are duplicates of long-standing requests
STEP 4: Generate the recap
Create a structured recap with these sections:
===DISCORD_START===
**Daily Issues Recap - ${TODAY}**
**Summary Stats**
- Total issues opened today: [count]
- By category: [bugs/features/questions]
**Critical/High Priority Issues**
[List any CRITICAL or HIGH severity issues with brief descriptions and issue numbers]
**Most Active/Discussed**
[Issues with significant engagement or from active community members]
**Trending Topics**
[Patterns noticed - e.g., 'Multiple reports about X', 'Continued interest in Y feature']
**Duplicates & Related**
[Issues that relate to existing open issues]
===DISCORD_END===
STEP 5: Format for Discord
Format the recap as a Discord-compatible message:
- Use Discord markdown (**, __, etc.)
- BE EXTREMELY CONCISE - this is an EOD summary, not a detailed report
- Use hyperlinked issue numbers with suppressed embeds: [#1234](<https://github.com/${{ github.repository }}/issues/1234>)
- Group related issues on single lines where possible
- Add emoji sparingly for critical items only
- HARD LIMIT: Keep under 1800 characters total
- Skip sections that have nothing notable (e.g., if no critical issues, omit that section)
- Prioritize signal over completeness - only surface what matters
OUTPUT: Output ONLY the content between ===DISCORD_START=== and ===DISCORD_END=== markers. Include the markers so I can extract it." > /tmp/recap_raw.txt
# Extract only the Discord message between markers
sed -n '/===DISCORD_START===/,/===DISCORD_END===/p' /tmp/recap_raw.txt | grep -v '===DISCORD' > /tmp/recap.txt
IMPORTANT: When counting comments/activity, EXCLUDE these bot accounts:
- copilot-pull-request-reviewer
- github-actions
STEP 3: Identify what matters (ONLY from today's PRs)
**Bug Fixes From Today:**
- PRs with 'fix' or 'bug' in title created/updated today
- Small bug fixes (< 100 lines changed) that are easy to review
- Bug fixes from community contributors
**High Activity Today:**
- PRs with significant human comments today (excluding bots listed above)
- PRs with back-and-forth discussion today
**Quick Wins:**
- Small PRs (< 50 lines) that are approved or nearly approved
- PRs that just need a final review
STEP 4: Generate the recap
Create a structured recap:
===DISCORD_START===
**Daily PR Recap - ${TODAY}**
**New PRs Today**
[PRs opened today - group by type: bug fixes, features, etc.]
**Active PRs Today**
[PRs with activity/updates today - significant discussion]
**Quick Wins**
[Small PRs ready to merge]
===DISCORD_END===
STEP 5: Format for Discord
- Use Discord markdown (**, __, etc.)
- BE EXTREMELY CONCISE - surface what we might miss
- Use hyperlinked PR numbers with suppressed embeds: [#1234](<https://github.com/${{ github.repository }}/pull/1234>)
- Include PR author: [#1234](<url>) (@author)
- For bug fixes, add brief description of what it fixes
- Show line count for quick wins: \"(+15/-3 lines)\"
- HARD LIMIT: Keep under 1800 characters total
- Skip empty sections
- Focus on PRs that need human eyes
OUTPUT: Output ONLY the content between ===DISCORD_START=== and ===DISCORD_END=== markers. Include the markers so I can extract it." > /tmp/pr_recap_raw.txt
# Extract only the Discord message between markers
sed -n '/===DISCORD_START===/,/===DISCORD_END===/p' /tmp/pr_recap_raw.txt | grep -v '===DISCORD' > /tmp/pr_recap.txt
opencode run --agent docs --model opencode/gpt-5.3-codex <<'EOF'
Update localized docs to match the latest English docs changes.
Changed English doc files:
<changed_english_docs>
${{ steps.changes.outputs.files }}
</changed_english_docs>
Requirements:
1. Update all relevant locale docs under packages/web/src/content/docs/<locale>/ so they reflect these English page changes.
2. You MUST use the Task tool for translation work and launch subagents with subagent_type `translator` (defined in .opencode/agent/translator.md).
3. Do not translate directly in the primary agent. Use translator subagent output as the source for locale text updates.
4. Run translator subagent Task calls in parallel whenever file/locale translation work is independent.
5. Use only the minimum tools needed for this task (read/glob, file edits, and translator Task). Do not use shell, web, search, or GitHub tools for translation work.
6. Preserve frontmatter keys, internal links, code blocks, and existing locale-specific metadata unless the English change requires an update.
7. Keep locale docs structure aligned with their corresponding English pages.
8. Do not modify English source docs in packages/web/src/content/docs/*.mdx.
9. If no locale updates are needed, make no changes.
EOF
- name:Commit and push locale docs updates
if:steps.changes.outputs.has_changes == 'true'
run:|
if [ -z "$(git status --porcelain)" ]; then
echo "No locale docs changes to commit"
exit 0
fi
git add -A
git commit -m "docs(i18n): sync locale docs from english changes"
opencode run -m opencode/claude-sonnet-4-6 "A new issue has been created:
Issue number: ${{ github.event.issue.number }}
Lookup this issue with gh issue view ${{ github.event.issue.number }}.
You have TWO tasks. Perform both, then post a SINGLE comment (if needed).
---
TASK 1: CONTRIBUTING GUIDELINES COMPLIANCE CHECK
Check whether the issue follows our contributing guidelines and issue templates.
This project has three issue templates that every issue MUST use one of:
1. Bug Report - requires a Description field with real content
2. Feature Request - requires a verification checkbox and description, title should start with [FEATURE]:
3. Question - requires the Question field with real content
Additionally check:
- No AI-generated walls of text (long, AI-generated descriptions are not acceptable)
- The issue has real content, not just template placeholder text left unchanged
- Bug reports should include some context about how to reproduce
- Feature requests should explain the problem or need
- We want to push for having the user provide system description & information
Do NOT be nitpicky about optional fields. Only flag real problems like: no template used, required fields empty or placeholder text only, obviously AI-generated walls of text, or completely empty/nonsensical content.
---
TASK 2: DUPLICATE CHECK
Search through existing issues (excluding #${{ github.event.issue.number }}) to find potential duplicates.
Consider:
1. Similar titles or descriptions
2. Same error messages or symptoms
3. Related functionality or components
4. Similar feature requests
Additionally, if the issue mentions keybinds, keyboard shortcuts, or key bindings, note the pinned keybinds issue #4997.
---
POSTING YOUR COMMENT:
Based on your findings, post a SINGLE comment on issue #${{ github.event.issue.number }}. Build the comment as follows:
If the issue is NOT compliant, start the comment with:
<!-- issue-compliance -->
Then explain what needs to be fixed and that they have 2 hours to edit the issue before it is automatically closed. Also add the label needs:compliance to the issue using: gh issue edit ${{ github.event.issue.number }} --add-label needs:compliance
If duplicates were found, include a section about potential duplicates with links.
If the issue mentions keybinds/keyboard shortcuts, include a note about #4997.
If the issue IS compliant AND no duplicates were found AND no keybind reference, do NOT comment at all.
Use this format for the comment:
[If not compliant:]
<!-- issue-compliance -->
This issue doesn't fully meet our [contributing guidelines](../blob/dev/CONTRIBUTING.md).
**What needs to be fixed:**
- [specific reasons]
Please edit this issue to address the above within **2 hours**, or it will be automatically closed.
[If duplicates found, add:]
---
This issue might be a duplicate of existing issues. Please check:
- #[issue_number]: [brief description of similarity]
[If keybind-related, add:]
For keybind-related issues, please also check our pinned keybinds documentation: #4997
[End with if not compliant:]
If you believe this was flagged incorrectly, please let a maintainer know.
Remember: post at most ONE comment combining all findings. If everything is fine, post nothing."
opencode run -m opencode/claude-sonnet-4-6 "Issue #${{ github.event.issue.number }} was previously flagged as non-compliant and has been edited.
Lookup this issue with gh issue view ${{ github.event.issue.number }}.
Re-check whether the issue now follows our contributing guidelines and issue templates.
This project has three issue templates that every issue MUST use one of:
1. Bug Report - requires a Description field with real content
2. Feature Request - requires a verification checkbox and description, title should start with [FEATURE]:
3. Question - requires the Question field with real content
Additionally check:
- No AI-generated walls of text (long, AI-generated descriptions are not acceptable)
- The issue has real content, not just template placeholder text left unchanged
- Bug reports should include some context about how to reproduce
- Feature requests should explain the problem or need
- We want to push for having the user provide system description & information
Do NOT be nitpicky about optional fields. Only flag real problems like: no template used, required fields empty or placeholder text only, obviously AI-generated walls of text, or completely empty/nonsensical content.
opencode run -m opencode/gpt-5.5 --variant medium "A new pull request has been created: '${PR_TITLE}'
<pr-number>
${{ steps.pr-number.outputs.number }}
</pr-number>
<pr-description>
$PR_BODY
</pr-description>
Please check all the code changes in this pull request against the style guide, also look for any bugs if they exist. Diffs are important but make sure you read the entire file to get proper context. Make it clear the suggestions are merely suggestions and the human can decide what to do
When critiquing code against the style guide, be sure that the code is ACTUALLY in violation, don't complain about else statements if they already use early returns there. You may complain about excessive nesting though, regardless of else statement usage.
When critiquing code style don't be a zealot, we don't like "let" statements but sometimes they are the simplest option, if someone does a bunch of nesting with let, they should consider using iife (see packages/opencode/src/util.iife.ts)
Use the gh cli to create comments on the files for the violations. Try to leave the comment on the exact line number. If you have a suggested fix include it in a suggestion code block.
If you are writing suggested fixes, BE SURE THAT the change you are recommending is actually valid typescript, often I have seen missing closing "}" or other syntax errors.
Generally, write a comment instead of writing suggested change if you can help it.
You are a duplicate PR detection agent. When a PR is opened, your job is to search for potentially duplicate or related open PRs.
Use the github-pr-search tool to search for PRs that might be addressing the same issue or feature.
IMPORTANT: The input will contain a line `CURRENT_PR_NUMBER: NNNN`. This is the current PR number, you should not mark that the current PR as a duplicate of itself.
Search using keywords from the PR title and description. Try multiple searches with different relevant terms.
If you find potential duplicates:
- List them with their titles and URLs
- Briefly explain why they might be related
If no duplicates are found, say so clearly. BUT ONLY SAY "No duplicate PRs found" (don't say anything else if no dups)
description: "Bump AI sdk dependencies minor / patch versions only"
---
Please read @package.json and @packages/opencode/package.json.
Your job is to look into AI SDK dependencies, figure out if they have versions that can be upgraded (minor or patch versions ONLY no major ignore major changes).
I want a report of every dependency and the version that can be upgraded to.
What would be even better is if you can give me brief summary of the changes for each dep and a link to the changelog for each dependency, or at least some reference info so I can see what bugs were fixed or new features were added.
Consider using subagents for each dep to save your context window.
Here is a short list of some deps (please be comprehensive tho):
- "ai"
- "@ai-sdk/openai"
- "@ai-sdk/anthropic"
- "@openrouter/ai-sdk-provider"
- etc, etc
DO NOT upgrade the dependencies yet, just make a list of all dependencies and their versions that can be upgraded to minor or patch versions only.
- Only include sections that have at least one notable entry
- Keep one bullet per commit you keep
- Skip commits that are entirely internal, CI, tests, refactors, or otherwise not user-facing
- Start each bullet with a capital letter
- Prefer what changed for users over what code changed internally
- Do not copy raw commit prefixes like `fix:` or `feat:` or trailing PR numbers like `(#123)`
- Community attribution is deterministic: only preserve an existing `(@username)` suffix from the changelog input
- If an input bullet has no `(@username)` suffix, do not add one
- Never add a new `(@username)` suffix from `git show`, commit authors, names, or email addresses
- If no notable entries remain and there is no contributor block, write exactly `No notable changes.`
- If no notable entries remain but there is a contributor block, omit all release sections and return only the contributor block
- If the input contains `## Community Contributors Input`, append the block below that heading to the end of the final file verbatim
- Do not add, remove, rewrite, or reorder contributor names or commit titles in that block
- Do not derive the thank-you section from the main summary bullets
- Do not include the heading `## Community Contributors Input` in the final file
- Focus on writing the least words to get your point across - users will skim read the changelog, so we should be precise
**Importantly, the changelog is for users (who are at least slightly technical), they may use the TUI, Desktop, SDK, Plugins and so forth. Be thorough in understanding flow on effects may not be immediately apparent. e.g. a package upgrade looks internal but may patch a bug. Or a refactor may also stabilise some race condition that fixes bugs for users. The PR title/body + commit message will give you the authors context, usually containing the outcome not just technical detail**
description: Extract non-obvious learnings from session to AGENTS.md files to build codebase understanding
---
Analyze this session and extract non-obvious learnings to add to AGENTS.md files.
AGENTS.md files can exist at any directory level, not just the project root. When an agent reads a file, any AGENTS.md in parent directories are automatically loaded into the context of the tool read. Place learnings as close to the relevant code as possible:
Use this folder for locale-specific translation guidance that supplements `.opencode/agent/translator.md`.
The global glossary in `translator.md` remains the source of truth for shared do-not-translate terms (commands, code, paths, product names, etc.). These locale files capture community learnings about phrasing and terminology preferences.
## File Naming
- One file per locale
- Use lowercase locale slugs that match docs locales when possible (for example, `zh-cn.md`, `zh-tw.md`)
- If only language-level guidance exists, use the language code (for example, `fr.md`)
- Some repo locale slugs may be aliases/non-BCP47 for consistency (for example, `br` for Brazilian Portuguese / `pt-BR`)
## What To Put In A Locale File
- **Sources**: PRs/issues/discussions that motivated the guidance
- **Do Not Translate (Locale Additions)**: locale-specific terms or casing decisions
- **Preferred Terms**: recurring UI/docs words with preferred translations
- **Guidance**: tone, style, and consistency notes
- **Avoid** (optional): common literal translations or wording we should avoid
- If the repo uses a locale alias slug, document the alias in **Guidance** (for example, prose may mention `pt-BR` while config/examples use `br`)
Prefer guidance that is:
- Repeated across multiple docs/screens
- Easy to apply consistently
- Backed by a community contribution or review discussion
description: Work with Effect v4 / effect-smol TypeScript code in this repo
---
# Effect
This codebase uses Effect for typed, composable TypeScript services, schemas, and workflows.
## Source Of Truth
Use the current Effect v4 / effect-smol source, not memory or older Effect v2/v3 examples.
1. If `.opencode/references/effect-smol` is missing, clone `https://github.com/Effect-TS/effect-smol` there. Do this in the project, not in the skill folder.
2. Search `.opencode/references/effect-smol` for exact APIs, examples, tests, and naming patterns before answering or implementing Effect-specific code.
3. Also inspect existing repo code for local house style before introducing new patterns.
4. Prefer answers and implementations backed by specific source files or nearby repo examples.
## Guidelines
- Prefer current Effect v4 APIs and project-local patterns over old blog posts, examples, or package-memory guesses.
- Use `Effect.gen(function* () { ... })` for multi-step workflows.
- Use `Effect.fn("Name")` or `Effect.fnUntraced(...)` for named effects when adding reusable service methods or important workflows.
- Prefer Effect `Schema` for API and domain data shapes. Use branded schemas for IDs and `Schema.TaggedErrorClass` for typed domain errors when modeling new error surfaces.
- Keep HTTP handlers thin: decode input, read request context, call services, and map transport errors. Put business rules in services.
- In Effect service code, prefer Effect-aware platform abstractions and dependencies over ad hoc promises where the surrounding code already does so.
- Keep layer composition explicit. Avoid broad hidden provisioning that makes missing dependencies hard to see.
- In tests, prefer the repo's existing Effect test helpers and live tests for filesystem, git, child process, locks, or timing behavior.
- Do not introduce `any`, non-null assertions, unchecked casts, or older Effect APIs just to satisfy types.
- Do not answer from memory. Verify against `.opencode/references/effect-smol` or nearby code first.
- To regenerate the JavaScript SDK, run `./packages/sdk/js/script/build.ts`.
- ALWAYS USE PARALLEL TOOLS WHEN APPLICABLE.
- The default branch in this repo is `dev`.
- Local `main` ref may not exist; use `dev` or `origin/dev` for diffs.
- Prefer automation: execute requested actions without confirmation unless blocked by missing info or safety/irreversibility.
## Style Guide
### General Principles
- Keep things in one function unless composable or reusable
- Avoid `try`/`catch` where possible
- Avoid using the `any` type
- Use Bun APIs when possible, like `Bun.file()`
- Rely on type inference when possible; avoid explicit type annotations or interfaces unless necessary for exports or clarity
- Prefer functional array methods (flatMap, filter, map) over for loops; use type guards on filter to maintain type inference downstream
- In `src/config`, follow the existing self-export pattern at the top of the file (for example `export * as ConfigAgent from "./agent"`) when adding a new config module.
Reduce total variable count by inlining when a value is only used once.
Replace `<platform>` with your platform (e.g., `darwin-arm64`, `linux-x64`).
- Core pieces:
- `packages/opencode`: OpenCode core business logic & server.
- `packages/opencode/src/cli/cmd/tui/`: The TUI code, written in SolidJS with [opentui](https://github.com/sst/opentui)
- `packages/app`: The shared web UI components, written in SolidJS
- `packages/desktop`: The native desktop app, built with Tauri (wraps `packages/app`)
- `packages/plugin`: Source for `@opencode-ai/plugin`
### Understanding bun dev vs opencode
During development, `bun dev` is the local equivalent of the built `opencode` command. Both run the same CLI interface:
```bash
# Development (from project root)
bun dev --help # Show all available commands
bun dev serve # Start headless API server
bun dev web # Start server + open web interface
bun dev <directory> # Start TUI in specific directory
# Production
opencode --help # Show all available commands
opencode serve # Start headless API server
opencode web # Start server + open web interface
opencode <directory> # Start TUI in specific directory
```
### Running the API Server
To start the OpenCode headless API server:
```bash
bun dev serve
```
This starts the headless server on port 4096 by default. You can specify a different port:
```bash
bun dev serve --port 8080
```
### Running the Web App
To test UI changes during development:
1. **First, start the OpenCode server** (see [Running the API Server](#running-the-api-server) section above)
2. **Then run the web app:**
```bash
bun run --cwd packages/app dev
```
This starts a local dev server at http://localhost:5173 (or similar port shown in output). Most UI changes can be tested here, but the server must be running for full functionality.
### Running the Desktop App
The desktop app is a native Tauri application that wraps the web UI.
To run the native desktop app:
```bash
bun run --cwd packages/desktop tauri dev
```
This starts the web dev server on http://localhost:1420 and opens the native window.
If you only want the web dev server (no native shell):
```bash
bun run --cwd packages/desktop dev
```
To create a production `dist/` and build the native app bundle:
```bash
bun run --cwd packages/desktop tauri build
```
This runs `bun run --cwd packages/desktop build` automatically via Tauri’s `beforeBuildCommand`.
> [!NOTE]
> Running the desktop app requires additional Tauri dependencies (Rust toolchain, platform-specific libraries). See the [Tauri prerequisites](https://v2.tauri.app/start/prerequisites/) for setup instructions.
> [!NOTE]
> If you make changes to the API or SDK (e.g. `packages/opencode/src/server/server.ts`), run `./script/generate.ts` to regenerate the SDK and related files.
Please try to follow the [style guide](./AGENTS.md)
### Setting up a Debugger
Bun debugging is currently rough around the edges. We hope this guide helps you get set up and avoid some pain points.
The most reliable way to debug OpenCode is to run it manually in a terminal via `bun run --inspect=<url> dev ...` and attach
your debugger via that URL. Other methods can result in breakpoints being mapped incorrectly, at least in VSCode (YMMV).
Caveats:
- If you want to run the OpenCode TUI and have breakpoints triggered in the server code, you might need to run `bun dev spawn` instead of
the usual `bun dev`. This is because `bun dev` runs the server in a worker thread and breakpoints might not work there.
- If `spawn` does not work for you, you can debug the server separately:
then attach TUI with `opencode attach http://localhost:4096`
- Debug TUI: `bun run --inspect=ws://localhost:6499/ --cwd packages/opencode --conditions=browser ./src/index.ts`
Other tips and tricks:
- You might want to use `--inspect-wait` or `--inspect-brk` instead of `--inspect`, depending on your workflow
- Specifying `--inspect=ws://localhost:6499/` on every invocation can be tiresome, you may want to `export BUN_OPTIONS=--inspect=ws://localhost:6499/` instead
#### VSCode Setup
If you use VSCode, you can use our example configurations [.vscode/settings.example.json](.vscode/settings.example.json) and [.vscode/launch.example.json](.vscode/launch.example.json).
Some debug methods that can be problematic:
- Debug configurations with `"request": "launch"` can have breakpoints incorrectly mapped and thus unusable
- The same problem arises when running OpenCode in the VSCode `JavaScript Debug Terminal`
With that said, you may want to try these methods, as they might work for you.
## Pull Request Expectations
### Issue First Policy
**All PRs must reference an existing issue.** Before opening a PR, open an issue describing the bug or feature. This helps maintainers triage and prevents duplicate work. PRs without a linked issue may be closed without review.
- Use `Fixes #123` or `Closes #123` in your PR description to link the issue
- For small fixes, a brief issue is fine - just enough context for maintainers to understand the problem
### General Requirements
- Keep pull requests small and focused
- Explain the issue and why your change fixes it
- Before adding new functionality, ensure it doesn't already exist elsewhere in the codebase
### UI Changes
If your PR includes UI changes, please include screenshots or videos showing the before and after. This helps maintainers review faster and gives you quicker feedback.
### Logic Changes
For non-UI changes (bug fixes, new features, refactors), explain **how you verified it works**:
- What did you test?
- How can a reviewer reproduce/confirm the fix?
### No AI-Generated Walls of Text
Long, AI-generated PR descriptions and issues are not acceptable and may be ignored. Respect the maintainers' time:
- Write short, focused descriptions
- Explain what changed and why in your own words
- If you can't explain it briefly, your PR might be too large
### PR Titles
PR titles should follow conventional commit standards:
- `feat:` new feature or functionality
- `fix:` bug fix
- `docs:` documentation or README changes
- `chore:` maintenance tasks, dependency updates, etc.
- `refactor:` code refactoring without changing behavior
- `test:` adding or updating tests
You can optionally include a scope to indicate which package is affected:
- `feat(app):` feature in the app package
- `fix(desktop):` bug fix in the desktop package
- `chore(opencode):` maintenance in the opencode package
Examples:
- `docs: update contributing guidelines`
- `fix: resolve crash on startup`
- `feat: add dark mode support`
- `feat(app): add dark mode support`
- `fix(desktop): resolve crash on startup`
- `chore: bump dependency versions`
### Style Preferences
These are not strictly enforced, they are just general guidelines:
- **Functions:** Keep logic within a single function unless breaking it out adds clear reuse or composition benefits.
- **Destructuring:** Do not do unnecessary destructuring of variables.
- **Control flow:** Avoid `else` statements.
- **Error handling:** Prefer `.catch(...)` instead of `try`/`catch` when possible.
- **Types:** Reach for precise types and avoid `any`.
- **Variables:** Stick to immutable patterns and avoid `let`.
- **Naming:** Choose concise single-word identifiers when they remain descriptive.
- **Runtime APIs:** Use Bun helpers such as `Bun.file()` when they fit the use case.
## Feature Requests
For net-new functionality, start with a design conversation. Open an issue describing the problem, your proposed approach (optional), and why it belongs in OpenCode. The core team will help decide whether it should move forward; please wait for that approval instead of opening a feature PR directly.
## Trust & Vouch System
This project uses [vouch](https://github.com/mitchellh/vouch) to manage contributor trust. The vouch list is maintained in [`.github/VOUCHED.td`](.github/VOUCHED.td).
### How it works
- **Vouched users** are explicitly trusted contributors.
- **Denounced users** are explicitly blocked. Issues and pull requests from denounced users are automatically closed. If you have been denounced, you can request to be unvouched by reaching out to a maintainer on [Discord](https://opencode.ai/discord)
- **Everyone else** can participate normally — you don't need to be vouched to open issues or PRs.
### For maintainers
Collaborators with write access can manage the vouch list by commenting on any issue:
- `vouch` — vouch for the issue author
- `vouch @username` — vouch for a specific user
- `denounce` — denounce the issue author
- `denounce @username` — denounce a specific user
- `denounce @username <reason>` — denounce with a reason
- `unvouch` / `unvouch @username` — remove someone from the list
Changes are committed automatically to `.github/VOUCHED.td`.
### Denouncement policy
Denouncement is reserved for users who repeatedly submit low-quality AI-generated contributions, spam, or otherwise act in bad faith. It is not used for disagreements or honest mistakes.
## Issue Requirements
All issues **must** use one of our issue templates:
- **Bug report** — for reporting bugs (requires a description)
- **Feature request** — for suggesting enhancements (requires verification checkbox and description)
- **Question** — for asking questions (requires the question)
Blank issues are not allowed. When a new issue is opened, an automated check verifies that it follows a template and meets our contributing guidelines. If an issue doesn't meet the requirements, you'll receive a comment explaining what needs to be fixed and have **2 hours** to edit the issue. After that, it will be automatically closed.
Issues may be flagged for:
- Not using a template
- Required fields left empty or filled with placeholder text
- AI-generated walls of text
- Missing meaningful content
If you believe your issue was incorrectly flagged, let a maintainer know.
# Handoff: Engram EL write-path field corruption + silent writes
**For:** Will (backend / EL soul)
**From:** Tim (via Claude Code)
**Date:** 2026-06-08
**Status:** Root cause confirmed; source fixes applied locally (NOT built/deployed); data analyzed; prune proposed (NOT applied).
---
## TL;DR
The EL wrapper `engram_node_full` had a **stale signature** that didn't match the C primitive. Because `el_val_t` is an untyped machine word, the compiler coerced caller args to the wrong declared types and forwarded them **by position** into a C function whose positions mean different things → `tier` got ints, `importance/confidence` got strings, `label` got a float, etc. One caller (`chat.el`) also put a *tier* into the `node_type` slot.
Source fixes are done. **You need to:** review, build with `elc`, restart the soul, verify, and apply the prune (daemon stopped). Details below.
**Coercion mechanism:**`el_val_t` is `uintptr_t` (`#define EL_STR(s) ((el_val_t)(uintptr_t)(s))`, `EL_INT(v) (v)`). The EL compiler binds each caller arg to the wrapper's *declared* param type (String→Float / String→Int coercion at the boundary), then the wrapper forwards **positionally**. Result for a correct-order caller `(content,"Memory","memory:remembered",sal,imp,conf,tier,tags)`:
-`label` ← `sal` (a float)
-`importance` ← a String
-`confidence` ← a String
-`tier` ← `ts` (the tier String coerced to Int) → **tier becomes an integer**
## 4. Fix applied — validation (defense in depth, `engram.el`)
Added `engram_valid_node_type` / `engram_valid_tier` allowlists. Both `engram_node` and `engram_node_full` now **reject invalid values with `__println` + return `""`** (fail loud, never silently write a malformed node).
- node_type allowlist: Memory, Knowledge, Belief, Project, Tag, BacklogItem, Artifact, Conversation, ExecutionContext, InternalStateEvent, Self, Entity, Process, ConfigEntry, Concept, Imprint *(union of the spec list + types actually present in the store — trim if some are illegitimate).*
- **Note:** `el_val_t` is untyped, so this catches wrong VALUES, not wrong TYPES. Type safety comes from the corrected signatures.
> All edits above are in the working tree on Tim's machine but **NOT compiled/deployed** and **NOT compile-verified** (no `elc` on that box).
---
## 5. DEPLOY RUNBOOK (your build env)
1. Pull the edited files: `el/lang/runtime/engram.el`, `neuron/chat.el`.
2. Build: `elc` (entry `neuron/soul.el`, import chain) → `neuron/dist/*.c`, then link as in `el/lang/install.sh` (`$(CC) $(CFLAGS) -o dist/neuron-fresh dist/*.c .../el_runtime.c -lcurl -lpthread`). Confirm `engram.el` recompiles into the import chain.
3. Restart the soul. **Note:** on Tim's box it's run by `/tmp/soul-keepalive.sh` (an auto-restart loop) → stop that loop before killing `neuron-fresh`, or it'll respawn the old binary.
4.**Verify (prove end-to-end):** write a node via the live API (POST `/api/memories` or the remember path) with an obvious throwaway label, then read it back and confirm `node_type` + `tier` are correct AND that it persisted (node_count increments; survives a snapshot save). There is **no delete endpoint** — clean up via the snapshot.
---
## 6. Data analysis + prune proposal (NOT applied)
- **~107 corrupt nodes** (node_type/tier not in the valid sets). node_type junk values: `''`, `'1'`, `'2'`, `'ntn-genesis'`, `'claude-opus-4-8'`, binary. tier junk: same + `'/Users/timlingo'`.
- **0 are field-repairable.** They're all genesis-bootstrap / binary detritus where *every* field (id/label/tier/tags) is corrupted together — 69× "You are ntn-genesis, a CGI.", 62× "ntn-genesis", ~70 binary garbage, plus a proxy URL + an API path that leaked into labels. No signal to reconstruct → **prune, don't fabricate.**
- **Proposal:** `~/.neuron/engram/snapshot.pruned.json` — 3,631 clean nodes (107 junk removed), edges intact (no dangling). Byte-verified: no *clean* node contains binary content, so re-encoding is lossless.
- **NOT applied** because the live daemon is **actively rewriting `snapshot.json`** (two reads returned different counts). Applying requires stopping the soul + keepalive, swapping in the pruned snapshot, then restarting. Do this in your controlled env with the backup retained.
---
## 7. Security heads-up (please action)
-`ANTHROPIC_API_KEY` is stored **in plaintext** in `/tmp/soul-keepalive.sh` — rotate it and move to a secret store.
- Internal infra leaked into node fields (`http://localhost:7771`, `/api/graph/edges?limit=5000`) — symptom of the same write bug; the prune removes those nodes.
## 8. Backlog of related gaps (separate from this fix)
- Soul chat loop reports **no tools** (`NONE`) / `NO_SHELL` — it narrates `curl`/`sqlite3` without executing. The capture REST path works, but the chat agent can't call it.
- **No `PUT`/`DELETE`** on knowledge nodes (`method not allowed`) — needed for UI edit/delete.
- No **source-conversation** edge on captured nodes — blocks "see source chat" in the UI.
- Writes have been **frozen since ~2026-04-29** (newest knowledge node) — nothing is being added in the current running state.
---
## ADDENDUM — Phase 0 live runtime findings (2026-06-08, verified against the running system)
Validated the write path end-to-end against `neuron-fresh :7770` + `engram :8742`. Confirms the diagnosis and corrects two common assumptions.
1.**`/api/neuron/knowledge/capture` + memory remember** — handled **in-process by the soul** (`neuron-api.el``handle_api_capture_knowledge` / remember → `engram_node_full(...)`). Live test: `POST …/knowledge/capture` returned `{"id":"2ccfc147…","ok":true}` but that id is **absent from `/api/graph/nodes` and `snapshot.json`** → the node corrupted/vanished. **This is exactly the `engram_node_full` wrapper bug this PR fixes.** It is NOT a `:7771` issue. → fixed by el PR #52 + soul rebuild.
2.**`/api/backlog`, `/api/memories`, `/api/knowledge`, `/api/artifacts`, `/api/projects`, `/api/imprints`** — `routes.el` proxies these to **`axon`** via `axon_get`/`axon_post` (base `SOUL_AXON` or default **`http://localhost:7771`**). `axon` = **`protocols/axon`, an unbuilt Rust crate**, not running → "Failed to connect to localhost port 7771." → needs axon stood up (separate Rust workstream) OR routes repointed.
**Architecture clarifications (so nobody chases the wrong port again):**
- The soul runs in **file-snapshot mode** (no `ENGRAM_URL` in `/tmp/soul-keepalive.sh`) → it uses `~/.neuron/engram/snapshot.json`, **not `engram :8742` live**. So writing to `:8742` does NOT make data visible to the soul the app talks to.
-`engram :8742` is its own EL service (`engram/src/server.el`) with a **working CRUD API**: `POST/GET/DELETE /api/nodes`, `/api/edges`, `/api/save`, `/api/load`, `/api/activate`, `/api/search`. Verified create+delete (`{"ok":true}`). **But** its `route_create_node` only reads `content/node_type/salience` — **no label/tier/tags/metadata** — so it can't set `metadata.tier_source: canonical`.
- Minor EL bug in `engram/src/server.el route_create_node`: `if str_eq(node_type,""){ let node_type = "Memory" }`**shadows** (new local) instead of reassigning → the default never applies; same for `salience`. Worth fixing while in there.
**Verification plan (run after the soul rebuild lands):**
1.`POST /api/neuron/knowledge/capture {content,title,tier:canonical}` → capture the returned id.
2.`GET /api/neuron/knowledge/search?q=<term>` → confirm the node comes back with correct `node_type`/`metadata.tier_source`.
3. Confirm it survives a snapshot save (present in `snapshot.json`). Only then is the write "real."
4. Backlog: once `axon :7771` is up, repeat for `POST /api/backlog`.
**Net:** "make writes persist" needs (a) **this wrapper fix built into the soul** (capture) and (b) **`axon :7771` running** (backlog/artifacts/etc.). Neither was doable on Tim's box (no `elc`; `axon` is unbuilt Rust — out of scope per the no-Rust guardrail). No live writes/restarts were performed; engram probe node was created and deleted to verify the API.
brew install anomalyco/tap/opencode # macOS و Linux (موصى به، دائما محدث)
brew install opencode # macOS و Linux (صيغة brew الرسمية، تحديث اقل)
sudo pacman -S opencode # Arch Linux (Stable)
paru -S opencode-bin # Arch Linux (Latest from AUR)
mise use -g opencode # اي نظام
nix run nixpkgs#opencode # او github:anomalyco/opencode لاحدث فرع dev
```
> [!TIP]
> احذف الاصدارات الاقدم من 0.1.x قبل التثبيت.
### تطبيق سطح المكتب (BETA)
يتوفر OpenCode ايضا كتطبيق سطح مكتب. قم بالتنزيل مباشرة من [صفحة الاصدارات](https://github.com/anomalyco/opencode/releases) او من [opencode.ai/download](https://opencode.ai/download).
يتضمن OpenCode وكيليْن (Agents) مدمجين يمكنك التبديل بينهما باستخدام زر `Tab`.
- **build** - الافتراضي، وكيل بصلاحيات كاملة لاعمال التطوير
- **plan** - وكيل للقراءة فقط للتحليل واستكشاف الكود
- يرفض تعديل الملفات افتراضيا
- يطلب الاذن قبل تشغيل اوامر bash
- مثالي لاستكشاف قواعد كود غير مألوفة او لتخطيط التغييرات
بالاضافة الى ذلك يوجد وكيل فرعي **general** للبحث المعقد والمهام متعددة الخطوات.
يستخدم داخليا ويمكن استدعاؤه بكتابة `@general` في الرسائل.
تعرف على المزيد حول [agents](https://opencode.ai/docs/agents).
### التوثيق
لمزيد من المعلومات حول كيفية ضبط OpenCode، [**راجع التوثيق**](https://opencode.ai/docs).
### المساهمة
اذا كنت مهتما بالمساهمة في OpenCode، يرجى قراءة [contributing docs](./CONTRIBUTING.md) قبل ارسال pull request.
### البناء فوق OpenCode
اذا كنت تعمل على مشروع مرتبط بـ OpenCode ويستخدم "opencode" كجزء من اسمه (مثل "opencode-dashboard" او "opencode-mobile")، يرجى اضافة ملاحظة في README توضح انه ليس مبنيا بواسطة فريق OpenCode ولا يرتبط بنا بأي شكل.
### FAQ
#### ما الفرق عن Claude Code؟
هو مشابه جدا لـ Claude Code من حيث القدرات. هذه هي الفروقات الاساسية:
- 100% مفتوح المصدر
- غير مقترن بمزود معين. نوصي بالنماذج التي نوفرها عبر [OpenCode Zen](https://opencode.ai/zen)؛ لكن يمكن استخدام OpenCode مع Claude او OpenAI او Google او حتى نماذج محلية. مع تطور النماذج ستتقلص الفجوات وستنخفض الاسعار، لذا من المهم ان يكون مستقلا عن المزود.
- دعم LSP جاهز للاستخدام
- تركيز على TUI. تم بناء OpenCode بواسطة مستخدمي neovim ومنشئي [terminal.shop](https://terminal.shop)؛ وسندفع حدود ما هو ممكن داخل الطرفية.
- معمارية عميل/خادم. على سبيل المثال، يمكن تشغيل OpenCode على جهازك بينما تقوده عن بعد من تطبيق جوال. هذا يعني ان واجهة TUI هي واحدة فقط من العملاء الممكنين.
---
**انضم الى مجتمعنا** [Discord](https://discord.gg/opencode) | [X.com](https://x.com/opencode)
brew install anomalyco/tap/opencode # macOS and Linux (recommended, always up to date)
brew install opencode # macOS and Linux (official brew formula, updated less)
sudo pacman -S opencode # Arch Linux (Stable)
paru -S opencode-bin # Arch Linux (Latest from AUR)
mise use -g opencode # Any OS
nix run nixpkgs#opencode # or github:anomalyco/opencode for latest dev branch
```
> [!TIP]
> ইনস্টল করার আগে ০.১.x এর চেয়ে পুরোনো ভার্সনগুলো মুছে ফেলুন।
### ডেস্কটপ অ্যাপ (BETA)
OpenCode ডেস্কটপ অ্যাপ্লিকেশন হিসেবেও উপলব্ধ। সরাসরি [রিলিজ পেজ](https://github.com/anomalyco/opencode/releases) অথবা [opencode.ai/download](https://opencode.ai/download) থেকে ডাউনলোড করুন।
OpenCode এ দুটি বিল্ট-ইন এজেন্ট রয়েছে যা আপনি `Tab` কি(key) দিয়ে পরিবর্তন করতে পারবেন।
- **build** - ডিফল্ট, ডেভেলপমেন্টের কাজের জন্য সম্পূর্ণ অ্যাক্সেসযুক্ত এজেন্ট
- **plan** - বিশ্লেষণ এবং কোড এক্সপ্লোরেশনের জন্য রিড-ওনলি এজেন্ট
- ডিফল্টভাবে ফাইল এডিট করতে দেয় না
- ব্যাশ কমান্ড চালানোর আগে অনুমতি চায়
- অপরিচিত কোডবেস এক্সপ্লোর করা বা পরিবর্তনের পরিকল্পনা করার জন্য আদর্শ
এছাড়াও জটিল অনুসন্ধান এবং মাল্টিস্টেপ টাস্কের জন্য একটি **general** সাবএজেন্ট অন্তর্ভুক্ত রয়েছে।
এটি অভ্যন্তরীণভাবে ব্যবহৃত হয় এবং মেসেজে `@general` লিখে ব্যবহার করা যেতে পারে।
এজেন্টদের সম্পর্কে আরও জানুন: [docs](https://opencode.ai/docs/agents)।
### ডকুমেন্টেশন (Documentation)
কিভাবে OpenCode কনফিগার করবেন সে সম্পর্কে আরও তথ্যের জন্য, [**আমাদের ডকস দেখুন**](https://opencode.ai/docs)।
### অবদান (Contributing)
আপনি যদি OpenCode এ অবদান রাখতে চান, অনুগ্রহ করে একটি পুল রিকোয়েস্ট সাবমিট করার আগে আমাদের [কন্ট্রিবিউটিং ডকস](./CONTRIBUTING.md) পড়ে নিন।
### OpenCode এর উপর বিল্ডিং (Building on OpenCode)
আপনি যদি এমন প্রজেক্টে কাজ করেন যা OpenCode এর সাথে সম্পর্কিত এবং প্রজেক্টের নামের অংশ হিসেবে "opencode" ব্যবহার করেন, উদাহরণস্বরূপ "opencode-dashboard" বা "opencode-mobile", তবে দয়া করে আপনার README তে একটি নোট যোগ করে স্পষ্ট করুন যে এই প্রজেক্টটি OpenCode দল দ্বারা তৈরি হয়নি এবং আমাদের সাথে এর কোনো সরাসরি সম্পর্ক নেই।
### সচরাচর জিজ্ঞাসিত প্রশ্নাবলী (FAQ)
#### এটি ক্লড কোড (Claude Code) থেকে কীভাবে আলাদা?
ক্যাপাবিলিটির দিক থেকে এটি ক্লড কোডের (Claude Code) মতই। এখানে মূল পার্থক্যগুলো দেওয়া হলো:
- ১০০% ওপেন সোর্স
- কোনো প্রোভাইডারের সাথে আবদ্ধ নয়। যদিও আমরা [OpenCode Zen](https://opencode.ai/zen) এর মাধ্যমে মডেলসমূহ ব্যবহারের পরামর্শ দিই, OpenCode ক্লড (Claude), ওপেনএআই (OpenAI), গুগল (Google), অথবা লোকাল মডেলগুলোর সাথেও ব্যবহার করা যেতে পারে। যেমন যেমন মডেলগুলো উন্নত হবে, তাদের মধ্যকার পার্থক্য কমে আসবে এবং দামও কমবে, তাই প্রোভাইডার-অজ্ঞাস্টিক হওয়া খুবই গুরুত্বপূর্ণ।
- আউট-অফ-দ্য-বক্স LSP সাপোর্ট
- TUI এর উপর ফোকাস। OpenCode নিওভিম (neovim) ব্যবহারকারী এবং [terminal.shop](https://terminal.shop) এর নির্মাতাদের দ্বারা তৈরি; আমরা টার্মিনালে কী কী সম্ভব তার সীমাবদ্ধতা ছাড়িয়ে যাওয়ার চেষ্টা করছি।
- ক্লায়েন্ট/সার্ভার আর্কিটেকচার। এটি যেমন OpenCode কে আপনার কম্পিউটারে চালানোর সুযোগ দেয়, তেমনি আপনি মোবাইল অ্যাপ থেকে রিমোটলি এটি নিয়ন্ত্রণ করতে পারবেন, অর্থাৎ TUI ফ্রন্টএন্ড কেবল সম্ভাব্য ক্লায়েন্টগুলোর মধ্যে একটি।
---
**আমাদের কমিউনিটিতে যুক্ত হোন** [Discord](https://discord.gg/opencode) | [X.com](https://x.com/opencode)
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
will.anderson
Neuron
tim.lingo