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.
Tim Lingo
will.anderson