Compare commits

..

20 Commits

Author SHA1 Message Date
will.anderson 35c189759c feat(runtime): add engram_wm_*, engram_load_merge, http_serve_async — needed by soul CI
El SDK Release / build-and-release (push) Successful in 8m44s
2026-06-11 13:40:10 -05:00
will.anderson 5c94b8680d Merge stage into main: corruption fix, model passthrough, UTF-8 escaping
El SDK Release / build-and-release (push) Successful in 11m22s
2026-06-10 17:37:41 -05:00
will.anderson cebf3ded62 Merge dev into stage: corruption fix + model passthrough
El SDK CI - stage / build-and-test (push) Failing after 11m30s
2026-06-10 17:37:27 -05:00
will.anderson b83ecf52f9 Merge pull request 'fix(runtime): pass model through to the LLM API (+ UTF-8 JSON escaping)' (#53) from fix/llm-model-and-utf8 into stage
El SDK CI - stage / build-and-test (push) Successful in 8m26s
fix(runtime): pass model through to LLM API + UTF-8 JSON escaping
2026-06-10 22:01:51 +00:00
will.anderson 15ea584671 Merge pull request 'Fix engram_node_full field corruption + add validation' (#52) from fix/engram-node-full-field-corruption into dev
El SDK CI - dev / build-and-test (push) Successful in 7m59s
Fix engram_node_full field corruption + add validation (+ SessionSummary allowlist)
2026-06-10 22:01:41 +00:00
Tim Lingo c2afcbddf5 fix(engram): allow SessionSummary node_type in validation allowlist
El SDK CI - dev / build-and-test (pull_request) Successful in 3m47s
handle_api_consolidate writes a "SessionSummary" node, but engram_valid_node_type
omitted it — so once this validation ships, every consolidate() would be silently
REJECTED at the engram boundary. Add SessionSummary to the allowlist.

Found in Will's PR review of neuron #1 / el #52.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-10 06:26:25 -05:00
Tim Lingo dbf2c659d9 fix(runtime): pass model through to the LLM API instead of dropping it
El SDK CI - stage / build-and-test (pull_request) Failing after 12s
llm_call_system / llm_call accepted a model argument and discarded it:
they called llm_chain_call(system, user) with no model, and the legacy
ANTHROPIC_API_KEY fallback passed NULL to llm_provider_request, so every
non-agentic chat was pinned to LLM_DEFAULT_MODEL (claude-sonnet-4-5)
regardless of the caller's selection.

Thread model_pref through llm_chain_call: provider-chain entries still
honor their own NEURON_LLM_N_MODEL override and fall back to the
requested model otherwise; the legacy Anthropic path now uses the
requested model. NULL/empty preserves prior default behavior.

Effect: the soul's model selection (state soul_model / SOUL_LLM_MODEL,
e.g. claude-opus-4-8) now reaches api.anthropic.com. Previously the
chat response echoed the selected model in its label while the request
billed Sonnet 4.5.

Not built locally (no elc/cc toolchain on this checkout); needs stage CI.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-09 08:03:56 -05:00
Tim Lingo 2b8062c55f fix(runtime): handle multi-byte UTF-8 in JSON string escaping
Validate UTF-8 continuation bytes in jb_emit_escaped; pass valid
sequences through and escape orphaned/invalid start bytes as \u00xx.
Pre-existing change found uncommitted in the working tree; committed
here so it is reviewable rather than lost.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-09 08:02:46 -05:00
Tim Lingo dfe4e83ed1 Fix engram_node_full wrapper field corruption + add node_type/tier validation
El SDK Release / build-and-release (pull_request) Failing after 9s
The wrapper signature was stale and didn't match the C primitive
__engram_node_full(content, node_type, label, salience, importance, confidence, tier, tags).
Because el_val_t is an untyped machine word, the compiler coerced caller args to the
wrong declared param types and forwarded them BY POSITION — so tier received an int,
importance/confidence received strings, label received a float, etc. (~100 corrupt nodes).

- Correct the wrapper to match the C contract 1:1 (no coercion, no reorder).
- Add engram_valid_node_type / engram_valid_tier allowlists; engram_node and
  engram_node_full now reject invalid values with __println + return "" (fail loud,
  no silent malformed write).

See neuron repo: HANDOFF-engram-write-corruption.md for the full write-up + deploy runbook.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-08 16:13:43 -05:00
will.anderson 2ed6b26dde Merge pull request 'promote: stage → main (all elb linker fixes + ci-base rebuild)' (#42) from stage into main
El SDK Release / build-and-release (push) Successful in 6m28s
promote: stage → main (all elb linker fixes + ci-base rebuild)
2026-05-07 14:25:37 +00:00
will.anderson d8e9fd12f4 Merge pull request 'promote: dev → stage (all elb linker fixes)' (#41) from dev into stage
El SDK Release / build-and-release (pull_request) Successful in 3m51s
El SDK CI - stage / build-and-test (push) Successful in 4m11s
promote: dev → stage (all elb linker fixes)
2026-05-07 14:20:53 +00:00
will.anderson 8fa9c4ba20 Merge pull request 'promote: dev → stage (elb linker fixes)' (#38) from dev into stage
El SDK Release / build-and-release (pull_request) Failing after 1m2s
El SDK CI - stage / build-and-test (push) Successful in 3m56s
promote: dev → stage (elb linker fixes)
2026-05-07 08:11:38 +00:00
will.anderson 9c7bde47dc Merge pull request 'promote: dev → stage (elb gcc fix)' (#35) from dev into stage
El SDK Release / build-and-release (pull_request) Failing after 40s
El SDK CI - stage / build-and-test (push) Successful in 3m45s
promote: dev → stage (elb gcc fix)
2026-05-07 08:01:22 +00:00
will.anderson c0553459e1 Merge pull request 'promote: dev → stage (CI rebuild fix + ci-base refresh)' (#32) from dev into stage
El SDK Release / build-and-release (pull_request) Failing after 35s
El SDK CI - stage / build-and-test (push) Successful in 3m47s
promote: dev → stage (CI rebuild fix + ci-base refresh)
2026-05-07 07:50:27 +00:00
will.anderson fd208583fe Merge pull request 'promote: dev → stage (elb build fix)' (#28) from dev into stage
El SDK CI - stage / build-and-test (push) Successful in 3m51s
El SDK Release / build-and-release (pull_request) Failing after 38s
promote: dev → stage (elb build fix)
2026-05-07 02:46:27 +00:00
will.anderson 3e29fc43ab Merge pull request 'promote: dev → stage (__http_do_map_to_file)' (#25) from dev into stage
El SDK CI - stage / build-and-test (push) Successful in 3m44s
El SDK Release / build-and-release (pull_request) Failing after 47s
2026-05-07 02:14:30 +00:00
will.anderson 979a5677d5 Merge pull request 'promote: dev → stage (__-prefixed runtime fix)' (#22) from dev into stage
El SDK CI - stage / build-and-test (push) Successful in 3m48s
El SDK Release / build-and-release (pull_request) Failing after 1m4s
2026-05-07 01:48:32 +00:00
will.anderson 17b1aa0736 Merge pull request 'promote: dev → stage (return type fix)' (#19) from dev into stage
El SDK CI - stage / build-and-test (push) Failing after 4m1s
El SDK Release / build-and-release (pull_request) Failing after 42s
2026-05-07 01:12:18 +00:00
will.anderson f0c731d2db Merge pull request 'promote: dev → stage (runtime fix)' (#16) from dev into stage
El SDK CI - stage / build-and-test (push) Successful in 3m43s
El SDK Release / build-and-release (pull_request) Failing after 45s
2026-05-07 00:43:52 +00:00
will.anderson e7e0f7d3e5 Merge pull request 'promote: dev → stage' (#12) from dev into stage
El SDK CI - stage / build-and-test (push) Successful in 4m3s
El SDK Release / build-and-release (pull_request) Failing after 37s
2026-05-07 00:23:46 +00:00
15 changed files with 527 additions and 3838 deletions
-132
View File
@@ -1,132 +0,0 @@
name: Engram CI
on:
push:
branches:
- main
- dev
paths:
- 'engram/**'
workflow_dispatch:
jobs:
build:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Install build dependencies
run: |
apt-get update -qq
apt-get install -y gcc libcurl4-openssl-dev apt-transport-https ca-certificates
echo "deb [trusted=yes] https://packages.cloud.google.com/apt cloud-sdk main" \
> /etc/apt/sources.list.d/google-cloud-sdk.list
apt-get update -qq && apt-get install -y google-cloud-cli
- name: Download El SDK from Artifact Registry
env:
GCP_SA_KEY: ${{ secrets.GCP_SA_KEY }}
run: |
echo "${GCP_SA_KEY}" > /tmp/gcp-key.json
gcloud auth activate-service-account --key-file=/tmp/gcp-key.json
gcloud config set project neuron-785695
rm -rf /opt/el/dist /opt/el/runtime
mkdir -p /opt/el/dist/platform /opt/el/dist/bin /opt/el/runtime
get_latest() {
gcloud artifacts versions list \
--repository=foundation-dev \
--location=us-central1 \
--project=neuron-785695 \
--package="$1" \
--sort-by="~createTime" \
--limit=1 \
--format="value(name)" 2>/dev/null | awk -F/ '{print $NF}'
}
ELC_VER=$(get_latest el-elc)
ELB_VER=$(get_latest el-elb)
RC_VER=$(get_latest el-runtime-c)
RH_VER=$(get_latest el-runtime-h)
echo "Downloading elc@${ELC_VER} elb@${ELB_VER} runtime-c@${RC_VER} runtime-h@${RH_VER}"
gcloud artifacts generic download \
--repository=foundation-dev --location=us-central1 --project=neuron-785695 \
--package=el-elc --version="${ELC_VER}" \
--destination=/opt/el/dist/platform/
gcloud artifacts generic download \
--repository=foundation-dev --location=us-central1 --project=neuron-785695 \
--package=el-elb --version="${ELB_VER}" \
--destination=/opt/el/dist/bin/
gcloud artifacts generic download \
--repository=foundation-dev --location=us-central1 --project=neuron-785695 \
--package=el-runtime-c --version="${RC_VER}" \
--destination=/opt/el/runtime/
gcloud artifacts generic download \
--repository=foundation-dev --location=us-central1 --project=neuron-785695 \
--package=el-runtime-h --version="${RH_VER}" \
--destination=/opt/el/runtime/
mv /opt/el/dist/platform/elc* /opt/el/dist/platform/elc 2>/dev/null || true
mv /opt/el/dist/bin/elb* /opt/el/dist/bin/elb 2>/dev/null || true
mv /opt/el/runtime/el_runtime.c* /opt/el/runtime/el_runtime.c 2>/dev/null || true
mv /opt/el/runtime/el_runtime.h* /opt/el/runtime/el_runtime.h 2>/dev/null || true
chmod +x /opt/el/dist/platform/elc /opt/el/dist/bin/elb
echo "El SDK ready"
- name: Build engram binary (linux/amd64)
run: |
ELB=/opt/el/dist/bin/elb
ELC=/opt/el/dist/platform/elc
RUNTIME=/opt/el/runtime
# elb reads manifest.el from the working directory.
# engram/dist/engram.c is the pre-compiled C translation of src/server.el.
# elb compiles dist/engram.c + el_runtime.c → dist/engram binary.
cd engram
"$ELB" --elc="$ELC" --runtime="$RUNTIME"
ls -lh dist/engram
file dist/engram
- name: Smoke test
run: |
file engram/dist/engram
timeout 3 engram/dist/engram --help 2>&1 || true
echo "smoke test complete"
- name: Publish engram binary to Artifact Registry
if: github.event_name == 'push'
env:
GCP_SA_KEY: ${{ secrets.GCP_SA_KEY }}
run: |
VERSION="${GITHUB_SHA:0:8}"
gcloud artifacts generic upload \
--repository=foundation-dev \
--location=us-central1 \
--project=neuron-785695 \
--package=engram \
--version="${VERSION}" \
--source=engram/dist/engram
# Re-upload as "latest" — Artifact Registry generic artifacts don't
# support moving tags, so we upload again. The newest upload wins.
gcloud artifacts generic upload \
--repository=foundation-dev \
--location=us-central1 \
--project=neuron-785695 \
--package=engram \
--version="latest" \
--source=engram/dist/engram \
2>/dev/null || true
echo "Published engram@${VERSION} and engram@latest"
rm -f /tmp/gcp-key.json
BIN
View File
Binary file not shown.
+21 -752
View File
@@ -2,11 +2,6 @@
#include <stdlib.h>
#include "el_runtime.h"
el_val_t bm25_tokenize(el_val_t text);
el_val_t bm25_count_term(el_val_t term, el_val_t doc_tokens);
el_val_t bm25_score_doc(el_val_t doc_content, el_val_t query_tokens, el_val_t corpus_size, el_val_t avg_doc_len);
el_val_t bm25_search_json(el_val_t query, el_val_t limit);
el_val_t auto_link_content_node(el_val_t node_id, el_val_t content);
el_val_t parse_port(el_val_t bind);
el_val_t ok_json(void);
el_val_t err_json(el_val_t msg);
@@ -25,246 +20,16 @@ el_val_t route_create_edge(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neighbors(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_strengthen(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_forget(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_decay(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_export(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_reindex(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_save(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_load(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_health(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_session_begin(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_ctx(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_memory(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_knowledge_capture(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_knowledge_evolve(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_knowledge_promote(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_recall(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_graph(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_graph_link(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_list(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_consolidate(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_config(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_state_events(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_neuron_processes(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_events_next(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_events_ack(el_val_t method, el_val_t path, el_val_t body);
el_val_t route_bm25_search(el_val_t method, el_val_t path, el_val_t body);
el_val_t check_auth_ok(el_val_t method, el_val_t body);
el_val_t handle_request(el_val_t method, el_val_t path, el_val_t body);
el_val_t bind_str;
el_val_t port;
el_val_t data_dir;
el_val_t db_path;
el_val_t loaded;
el_val_t bm25_tokenize(el_val_t text) {
el_val_t t = str_to_lower(text);
t = str_replace(t, EL_STR("."), EL_STR(" "));
t = str_replace(t, EL_STR(","), EL_STR(" "));
t = str_replace(t, EL_STR("!"), EL_STR(" "));
t = str_replace(t, EL_STR("?"), EL_STR(" "));
t = str_replace(t, EL_STR("\""), EL_STR(" "));
t = str_replace(t, EL_STR(":"), EL_STR(" "));
t = str_replace(t, EL_STR(";"), EL_STR(" "));
t = str_replace(t, EL_STR("("), EL_STR(" "));
t = str_replace(t, EL_STR(")"), EL_STR(" "));
t = str_replace(t, EL_STR("["), EL_STR(" "));
t = str_replace(t, EL_STR("]"), EL_STR(" "));
t = str_replace(t, EL_STR("{"), EL_STR(" "));
t = str_replace(t, EL_STR("}"), EL_STR(" "));
t = str_replace(t, EL_STR("/"), EL_STR(" "));
t = str_replace(t, EL_STR("\\"), EL_STR(" "));
t = str_replace(t, EL_STR("'"), EL_STR(" "));
t = str_replace(t, EL_STR("-"), EL_STR(" "));
t = str_replace(t, EL_STR("_"), EL_STR(" "));
t = str_replace(t, EL_STR("+"), EL_STR(" "));
return str_trim(t);
return 0;
}
el_val_t bm25_count_term(el_val_t term, el_val_t doc_tokens) {
el_val_t padded_term = el_str_concat(el_str_concat(EL_STR(" "), term), EL_STR(" "));
el_val_t padded_doc = el_str_concat(el_str_concat(EL_STR(" "), doc_tokens), EL_STR(" "));
return str_count(padded_doc, padded_term);
return 0;
}
el_val_t bm25_score_doc(el_val_t doc_content, el_val_t query_tokens, el_val_t corpus_size, el_val_t avg_doc_len) {
el_val_t k1 = el_from_float(1.2);
el_val_t b = el_from_float(0.75);
el_val_t delta = el_from_float(1.0);
el_val_t doc_tokens = bm25_tokenize(doc_content);
el_val_t doc_wc = str_count_words(doc_tokens);
if (doc_wc == 0) {
return EL_STR("0.0");
}
el_val_t doc_len = int_to_float(doc_wc);
el_val_t avg_len = str_to_float(avg_doc_len);
el_val_t N = int_to_float(corpus_size);
el_val_t idf_arg = float_add(float_div(float_add(N, el_from_float(1.2)), el_from_float(1.5)), el_from_float(1.0));
el_val_t idf = math_log(idf_arg);
el_val_t terms = str_split(query_tokens, EL_STR(" "));
el_val_t n_terms = len(terms);
el_val_t score = el_from_float(0.0);
el_val_t i = 0;
while (i < n_terms) {
el_val_t term = get(terms, i);
el_val_t tlen = str_len(term);
if (tlen >= 2) {
el_val_t tf_count = bm25_count_term(term, doc_tokens);
if (tf_count > 0) {
el_val_t tf_raw = int_to_float(tf_count);
el_val_t norm_factor = float_add(float_sub(el_from_float(1.0), b), float_div(float_mul(b, doc_len), avg_len));
el_val_t numerator = float_mul(tf_raw, float_add(k1, el_from_float(1.0)));
el_val_t denominator = float_add(tf_raw, float_mul(k1, norm_factor));
el_val_t tf_comp = float_add(delta, float_div(numerator, denominator));
score = float_add(score, float_mul(idf, tf_comp));
}
}
i = (i + 1);
}
return float_to_str(score);
return 0;
}
el_val_t bm25_search_json(el_val_t query, el_val_t limit) {
el_val_t scan_limit = (limit * 10);
if (scan_limit < 200) {
scan_limit = 200;
}
if (scan_limit > 5000) {
scan_limit = 5000;
}
el_val_t nodes_json = engram_scan_nodes_json(scan_limit, 0);
el_val_t n = json_array_len(nodes_json);
if (n == 0) {
return EL_STR("[]");
}
el_val_t total_words = 0;
el_val_t i = 0;
while (i < n) {
el_val_t node = json_array_get(nodes_json, i);
el_val_t content = json_get_string(node, EL_STR("content"));
el_val_t tokens = bm25_tokenize(content);
el_val_t wc = str_count_words(tokens);
total_words = (total_words + wc);
i = (i + 1);
}
el_val_t avg_doc_len_f = float_div(int_to_float(total_words), int_to_float(n));
el_val_t avg_doc_len = ({ el_val_t _if_result_1 = 0; if (float_gt(avg_doc_len_f, el_from_float(0.0))) { _if_result_1 = (float_to_str(avg_doc_len_f)); } else { _if_result_1 = (EL_STR("1.0")); } _if_result_1; });
el_val_t query_tokens = bm25_tokenize(query);
if (str_eq(str_trim(query_tokens), EL_STR(""))) {
return EL_STR("[]");
}
el_val_t result_nodes = 0;
el_val_t result_scores = 0;
el_val_t result_count = 0;
el_val_t j = 0;
while (j < n) {
el_val_t node = json_array_get(nodes_json, j);
el_val_t content = json_get_string(node, EL_STR("content"));
el_val_t sc_str = bm25_score_doc(content, query_tokens, n, avg_doc_len);
if (float_gt(str_to_float(sc_str), el_from_float(0.0))) {
result_nodes = list_push(result_nodes, node);
result_scores = list_push(result_scores, sc_str);
result_count = (result_count + 1);
}
j = (j + 1);
}
if (result_count == 0) {
return EL_STR("[]");
}
el_val_t out_limit = ({ el_val_t _if_result_2 = 0; if ((result_count < limit)) { _if_result_2 = (result_count); } else { _if_result_2 = (limit); } _if_result_2; });
el_val_t k = 0;
while (k < out_limit) {
el_val_t max_idx = k;
el_val_t max_sc_str = get(result_scores, k);
el_val_t max_sc_f = str_to_float(max_sc_str);
el_val_t p = (k + 1);
while (p < result_count) {
el_val_t sc2_str = get(result_scores, p);
el_val_t sc2_f = str_to_float(sc2_str);
if (float_gt(sc2_f, max_sc_f)) {
max_sc_f = sc2_f;
max_sc_str = sc2_str;
max_idx = p;
}
p = (p + 1);
}
if (max_idx != k) {
el_val_t tmp_node = get(result_nodes, k);
el_val_t tmp_sc = get(result_scores, k);
result_nodes = list_set(result_nodes, k, get(result_nodes, max_idx));
result_scores = list_set(result_scores, k, get(result_scores, max_idx));
result_nodes = list_set(result_nodes, max_idx, tmp_node);
result_scores = list_set(result_scores, max_idx, tmp_sc);
}
k = (k + 1);
}
el_val_t out = EL_STR("[");
el_val_t r = 0;
while (r < out_limit) {
el_val_t node = get(result_nodes, r);
el_val_t sc_str = get(result_scores, r);
el_val_t node_len = str_len(node);
el_val_t node_body = str_slice(node, 0, (node_len - 1));
el_val_t entry = el_str_concat(el_str_concat(el_str_concat(node_body, EL_STR(",\"bm25_score\":")), sc_str), EL_STR("}"));
if (r > 0) {
out = el_str_concat(out, EL_STR(","));
}
out = el_str_concat(out, entry);
r = (r + 1);
}
return el_str_concat(out, EL_STR("]"));
return 0;
}
el_val_t auto_link_content_node(el_val_t node_id, el_val_t content) {
el_val_t clen = str_len(content);
if (clen < 20) {
return 0;
}
el_val_t sp1 = str_index_of(content, EL_STR(" "));
el_val_t w1end = ({ el_val_t _if_result_3 = 0; if ((sp1 < 0)) { _if_result_3 = (clen); } else { _if_result_3 = (sp1); } _if_result_3; });
el_val_t word1 = str_slice(content, 0, w1end);
state_set(EL_STR("aln_term"), EL_STR(""));
if (str_len(word1) >= 5) {
state_set(EL_STR("aln_term"), word1);
}
if (str_eq(state_get(EL_STR("aln_term")), EL_STR(""))) {
if (sp1 >= 0) {
el_val_t rest = str_slice(content, (sp1 + 1), clen);
el_val_t sp2 = str_index_of(rest, EL_STR(" "));
el_val_t w2end = ({ el_val_t _if_result_4 = 0; if ((sp2 < 0)) { _if_result_4 = (str_len(rest)); } else { _if_result_4 = (sp2); } _if_result_4; });
el_val_t word2 = str_slice(rest, 0, w2end);
if (str_len(word2) >= 5) {
state_set(EL_STR("aln_term"), word2);
}
}
}
el_val_t search_term = state_get(EL_STR("aln_term"));
if (str_eq(search_term, EL_STR(""))) {
return 0;
}
el_val_t results = bm25_search_json(search_term, 20);
el_val_t n = json_array_len(results);
state_set(EL_STR("aln_linked"), EL_STR("0"));
el_val_t i = 0;
while (i < n) {
el_val_t linked_so_far = str_to_int(state_get(EL_STR("aln_linked")));
if (linked_so_far < 3) {
el_val_t elem = json_array_get(results, i);
el_val_t rid = json_get_string(elem, EL_STR("id"));
el_val_t rtype = json_get_string(elem, EL_STR("node_type"));
if ((!str_eq(rtype, EL_STR("InternalStateEvent")) && !str_eq(rid, EL_STR(""))) && !str_eq(rid, node_id)) {
engram_connect(node_id, rid, el_from_float(0.6), EL_STR("related"));
state_set(EL_STR("aln_linked"), int_to_str((linked_so_far + 1)));
}
}
i = (i + 1);
}
return str_to_int(state_get(EL_STR("aln_linked")));
return 0;
}
el_val_t snapshot_path;
el_val_t parse_port(el_val_t bind) {
el_val_t colon = str_index_of(bind, EL_STR(":"));
@@ -354,8 +119,7 @@ el_val_t route_create_node(el_val_t method, el_val_t path, el_val_t body) {
salience = el_from_float(0.5);
}
el_val_t id = engram_node(content, node_type, salience);
el_val_t auto_linked = auto_link_content_node(id, content);
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"id\":\""), id), EL_STR("\",\"content\":\"")), content), EL_STR("\",\"node_type\":\"")), node_type), EL_STR("\",\"auto_linked\":")), int_to_str(auto_linked)), EL_STR("}"));
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"id\":\""), id), EL_STR("\",\"content\":\"")), content), EL_STR("\",\"node_type\":\"")), node_type), EL_STR("\"}"));
return 0;
}
@@ -412,7 +176,7 @@ el_val_t route_search(el_val_t method, el_val_t path, el_val_t body) {
if (limit == 0) {
limit = 20;
}
return bm25_search_json(q, limit);
return engram_search_json(q, limit);
return 0;
}
@@ -429,17 +193,6 @@ el_val_t route_activate(el_val_t method, el_val_t path, el_val_t body) {
depth = bd;
}
}
el_val_t top = bm25_search_json(q, 10);
el_val_t nb = json_array_len(top);
el_val_t bi = 0;
while (bi < nb) {
el_val_t node = json_array_get(top, bi);
el_val_t nid = json_get_string(node, EL_STR("id"));
if (!str_eq(nid, EL_STR(""))) {
engram_strengthen(nid);
}
bi = (bi + 1);
}
return el_str_concat(el_str_concat(EL_STR("{\"results\":"), engram_activate_json(q, depth)), EL_STR("}"));
return 0;
}
@@ -490,46 +243,30 @@ el_val_t route_forget(el_val_t method, el_val_t path, el_val_t body) {
return 0;
}
el_val_t route_decay(el_val_t method, el_val_t path, el_val_t body) {
return engram_apply_decay_json();
return 0;
}
el_val_t route_export(el_val_t method, el_val_t path, el_val_t body) {
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
el_val_t db_path = el_str_concat(dir, EL_STR("/engram.db"));
engram_write_binary_el(db_path);
el_val_t route_save(el_val_t method, el_val_t path, el_val_t body) {
el_val_t p = json_get_string(body, EL_STR("path"));
if (str_eq(p, EL_STR(""))) {
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
p = el_str_concat(dir, EL_STR("/snapshot.json"));
}
engram_save(p);
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"binary\":\""), db_path), EL_STR("\",\"json\":\"")), p), EL_STR("\"}"));
return 0;
}
el_val_t route_reindex(el_val_t method, el_val_t path, el_val_t body) {
return engram_reindex_json();
return el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"path\":\""), p), EL_STR("\"}"));
return 0;
}
el_val_t route_load(el_val_t method, el_val_t path, el_val_t body) {
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
el_val_t db_path = el_str_concat(dir, EL_STR("/engram.db"));
el_val_t ok = engram_load_binary_el(db_path);
if (!ok) {
el_val_t p = json_get_string(body, EL_STR("path"));
if (str_eq(p, EL_STR(""))) {
p = el_str_concat(dir, EL_STR("/snapshot.json"));
el_val_t p = json_get_string(body, EL_STR("path"));
if (str_eq(p, EL_STR(""))) {
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
engram_load(p);
p = el_str_concat(dir, EL_STR("/snapshot.json"));
}
engram_load(p);
return ok_json();
return 0;
}
@@ -539,344 +276,6 @@ el_val_t route_health(el_val_t method, el_val_t path, el_val_t body) {
return 0;
}
el_val_t route_neuron_session_begin(el_val_t method, el_val_t path, el_val_t body) {
el_val_t results = engram_activate_json(EL_STR("memory knowledge context"), 2);
el_val_t nc = engram_node_count();
el_val_t ec = engram_edge_count();
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"nodes\":"), results), EL_STR(",\"node_count\":")), int_to_str(nc)), EL_STR(",\"edge_count\":")), int_to_str(ec)), EL_STR("}"));
return 0;
}
el_val_t route_neuron_ctx(el_val_t method, el_val_t path, el_val_t body) {
el_val_t results = engram_activate_json(EL_STR("architecture decision memory"), 2);
el_val_t n = json_array_len(results);
el_val_t limit = ({ el_val_t _if_result_5 = 0; if ((n > 10)) { _if_result_5 = (10); } else { _if_result_5 = (n); } _if_result_5; });
el_val_t ctx = EL_STR("Recent working memory:\n");
el_val_t i = 0;
el_val_t ctx_body = EL_STR("");
while (i < limit) {
el_val_t elem = json_array_get(results, i);
el_val_t label = json_get_string(elem, EL_STR("label"));
el_val_t content = json_get_string(elem, EL_STR("content"));
el_val_t clen = str_len(content);
el_val_t snippet = ({ el_val_t _if_result_6 = 0; if ((clen > 200)) { _if_result_6 = (str_slice(content, 0, 200)); } else { _if_result_6 = (content); } _if_result_6; });
ctx_body = el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(ctx_body, EL_STR("- [")), label), EL_STR("]: ")), snippet), EL_STR("\n"));
i = (i + 1);
}
el_val_t full_ctx = el_str_concat(ctx, ctx_body);
return el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"context\":\""), str_replace(str_replace(str_replace(full_ctx, EL_STR("\\"), EL_STR("\\\\")), EL_STR("\""), EL_STR("\\\"")), EL_STR("\n"), EL_STR("\\n"))), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_memory(el_val_t method, el_val_t path, el_val_t body) {
el_val_t content = json_get_string(body, EL_STR("content"));
if (str_eq(content, EL_STR(""))) {
return EL_STR("{\"error\":\"content is required\"}");
}
el_val_t node_type = json_get_string(body, EL_STR("node_type"));
if (str_eq(node_type, EL_STR(""))) {
node_type = EL_STR("Memory");
}
el_val_t label = json_get_string(body, EL_STR("label"));
el_val_t importance = json_get_string(body, EL_STR("importance"));
el_val_t project = json_get_string(body, EL_STR("project"));
el_val_t tags_raw = json_get_string(body, EL_STR("tags"));
el_val_t tier = EL_STR("Episodic");
if (str_eq(importance, EL_STR("critical"))) {
tier = EL_STR("Procedural");
}
if (str_eq(importance, EL_STR("high"))) {
tier = EL_STR("Semantic");
}
if (str_eq(importance, EL_STR("normal"))) {
tier = EL_STR("Episodic");
}
if (str_eq(importance, EL_STR("low"))) {
tier = EL_STR("Working");
}
el_val_t explicit_tier = json_get_string(body, EL_STR("tier"));
if (!str_eq(explicit_tier, EL_STR(""))) {
tier = explicit_tier;
}
el_val_t tags_str = tags_raw;
if (!str_eq(project, EL_STR(""))) {
if (str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(EL_STR("project:"), project);
}
if (!str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(el_str_concat(tags_str, EL_STR(" project:")), project);
}
}
el_val_t id = engram_node_full(content, node_type, label, el_from_float(0.5), el_from_float(0.5), el_from_float(1.0), tier, tags_str);
el_val_t auto_linked = auto_link_content_node(id, content);
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
el_val_t db_path = el_str_concat(dir, EL_STR("/engram.db"));
engram_write_binary_el(db_path);
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"id\":\""), id), EL_STR("\",\"auto_linked\":")), int_to_str(auto_linked)), EL_STR(",\"content\":\"")), str_replace(str_replace(content, EL_STR("\\"), EL_STR("\\\\")), EL_STR("\""), EL_STR("\\\""))), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_knowledge_capture(el_val_t method, el_val_t path, el_val_t body) {
el_val_t content = json_get_string(body, EL_STR("content"));
if (str_eq(content, EL_STR(""))) {
return EL_STR("{\"error\":\"content is required\"}");
}
el_val_t title = json_get_string(body, EL_STR("title"));
el_val_t category = json_get_string(body, EL_STR("category"));
el_val_t tags_raw = json_get_string(body, EL_STR("tags"));
el_val_t project = json_get_string(body, EL_STR("project"));
el_val_t tier_raw = json_get_string(body, EL_STR("tier"));
el_val_t tier = EL_STR("Episodic");
if (str_eq(tier_raw, EL_STR("lesson"))) {
tier = EL_STR("Semantic");
}
if (str_eq(tier_raw, EL_STR("canonical"))) {
tier = EL_STR("Procedural");
}
if (str_eq(tier_raw, EL_STR("note"))) {
tier = EL_STR("Episodic");
}
el_val_t tags_str = tags_raw;
if (!str_eq(category, EL_STR(""))) {
if (str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(EL_STR("category:"), category);
}
if (!str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(el_str_concat(tags_str, EL_STR(" category:")), category);
}
}
if (!str_eq(project, EL_STR(""))) {
if (str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(EL_STR("project:"), project);
}
if (!str_eq(tags_str, EL_STR(""))) {
tags_str = el_str_concat(el_str_concat(tags_str, EL_STR(" project:")), project);
}
}
el_val_t id = engram_node_full(content, EL_STR("Knowledge"), title, el_from_float(0.7), el_from_float(0.7), el_from_float(1.0), tier, tags_str);
el_val_t auto_linked = auto_link_content_node(id, content);
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
el_val_t db_path = el_str_concat(dir, EL_STR("/engram.db"));
engram_write_binary_el(db_path);
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"id\":\""), id), EL_STR("\",\"auto_linked\":")), int_to_str(auto_linked)), EL_STR("}"));
return 0;
}
el_val_t route_neuron_knowledge_evolve(el_val_t method, el_val_t path, el_val_t body) {
el_val_t content = json_get_string(body, EL_STR("content"));
el_val_t prior_id = json_get_string(body, EL_STR("id"));
if (str_eq(content, EL_STR(""))) {
return EL_STR("{\"ok\":true}");
}
el_val_t id = engram_node_full(content, EL_STR("Knowledge"), EL_STR(""), el_from_float(0.7), el_from_float(0.7), el_from_float(1.0), EL_STR("Semantic"), EL_STR("evolved"));
if (!str_eq(prior_id, EL_STR("")) && !str_eq(id, EL_STR(""))) {
engram_connect(id, prior_id, el_from_float(1.0), EL_STR("supersedes"));
}
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
engram_write_binary_el(el_str_concat(dir, EL_STR("/engram.db")));
return el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"id\":\""), id), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_knowledge_promote(el_val_t method, el_val_t path, el_val_t body) {
el_val_t id = json_get_string(body, EL_STR("id"));
if (str_eq(id, EL_STR(""))) {
return EL_STR("{\"ok\":true}");
}
el_val_t node_json = engram_get_node_json(id);
if (str_eq(node_json, EL_STR(""))) {
return err_json(EL_STR("node not found"));
}
if (str_eq(node_json, EL_STR("null"))) {
return err_json(EL_STR("node not found"));
}
el_val_t content = json_get_string(node_json, EL_STR("content"));
if (str_eq(content, EL_STR(""))) {
return err_json(EL_STR("node has no content"));
}
el_val_t label = json_get_string(node_json, EL_STR("label"));
el_val_t tags = json_get_string(node_json, EL_STR("tags"));
el_val_t current_tier = json_get_string(node_json, EL_STR("tier"));
el_val_t tier_raw = json_get_string(body, EL_STR("tier"));
el_val_t new_tier = EL_STR("");
if (str_eq(tier_raw, EL_STR("lesson"))) {
new_tier = EL_STR("Semantic");
}
if (str_eq(tier_raw, EL_STR("canonical"))) {
new_tier = EL_STR("Procedural");
}
if (str_eq(tier_raw, EL_STR("note"))) {
new_tier = EL_STR("Episodic");
}
if (str_eq(new_tier, EL_STR(""))) {
if (str_eq(current_tier, EL_STR("Working"))) {
new_tier = EL_STR("Episodic");
}
if (str_eq(current_tier, EL_STR("Episodic"))) {
new_tier = EL_STR("Semantic");
}
if (str_eq(current_tier, EL_STR("Semantic"))) {
new_tier = EL_STR("Procedural");
}
if (str_eq(current_tier, EL_STR("Procedural"))) {
new_tier = EL_STR("Procedural");
}
}
if (str_eq(new_tier, EL_STR(""))) {
new_tier = EL_STR("Semantic");
}
el_val_t new_id = engram_node_full(content, EL_STR("Knowledge"), label, el_from_float(0.7), el_from_float(0.8), el_from_float(1.0), new_tier, tags);
if (!str_eq(new_id, EL_STR(""))) {
engram_connect(new_id, id, el_from_float(1.0), EL_STR("supersedes"));
}
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
engram_write_binary_el(el_str_concat(dir, EL_STR("/engram.db")));
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"id\":\""), new_id), EL_STR("\",\"promoted_from\":\"")), id), EL_STR("\",\"tier\":\"")), new_tier), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_recall(el_val_t method, el_val_t path, el_val_t body) {
el_val_t query = json_get_string(body, EL_STR("query"));
el_val_t chain = json_get_string(body, EL_STR("chain_name"));
el_val_t limit = json_get_int(body, EL_STR("limit"));
if (limit == 0) {
limit = 20;
}
el_val_t q = ({ el_val_t _if_result_7 = 0; if (str_eq(query, EL_STR(""))) { _if_result_7 = (chain); } else { _if_result_7 = (query); } _if_result_7; });
if (str_eq(q, EL_STR(""))) {
return engram_scan_nodes_json(limit, 0);
}
return bm25_search_json(q, limit);
return 0;
}
el_val_t route_neuron_graph(el_val_t method, el_val_t path, el_val_t body) {
el_val_t id = query_param(path, EL_STR("id"));
if (str_eq(id, EL_STR(""))) {
return EL_STR("{\"error\":\"id is required\"}");
}
el_val_t node_json = engram_get_node_json(id);
return el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"node\":"), node_json), EL_STR(",\"neighbors\":[]}"));
return 0;
}
el_val_t route_neuron_graph_link(el_val_t method, el_val_t path, el_val_t body) {
el_val_t from_id = json_get_string(body, EL_STR("from_id"));
el_val_t to_id = json_get_string(body, EL_STR("to_id"));
if (str_eq(from_id, EL_STR("")) || str_eq(to_id, EL_STR(""))) {
return EL_STR("{\"error\":\"from_id and to_id are required\"}");
}
el_val_t relation = json_get_string(body, EL_STR("relation"));
if (str_eq(relation, EL_STR(""))) {
relation = EL_STR("related");
}
el_val_t weight = json_get_float(body, EL_STR("weight"));
if (str_eq(weight, el_from_float(0.0))) {
weight = el_from_float(0.5);
}
engram_connect(from_id, to_id, weight, relation);
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"from_id\":\""), from_id), EL_STR("\",\"to_id\":\"")), to_id), EL_STR("\",\"relation\":\"")), relation), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_list(el_val_t method, el_val_t path, el_val_t body) {
el_val_t clean = strip_query(path);
el_val_t prefix = EL_STR("/api/neuron/list/");
el_val_t node_type = str_slice(clean, str_len(prefix), str_len(clean));
el_val_t limit = query_int(path, EL_STR("limit"), 50);
if (str_eq(node_type, EL_STR(""))) {
return EL_STR("[]");
}
return engram_scan_nodes_by_type_json(node_type, limit, 0);
return 0;
}
el_val_t route_neuron_consolidate(el_val_t method, el_val_t path, el_val_t body) {
el_val_t dir = env(EL_STR("ENGRAM_DATA_DIR"));
if (str_eq(dir, EL_STR(""))) {
dir = EL_STR("/tmp/engram");
}
el_val_t db_path = el_str_concat(dir, EL_STR("/engram.db"));
engram_write_binary_el(db_path);
el_val_t nc = engram_node_count();
el_val_t ec = engram_edge_count();
return el_str_concat(el_str_concat(el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"node_count\":"), int_to_str(nc)), EL_STR(",\"edge_count\":")), int_to_str(ec)), EL_STR("}"));
return 0;
}
el_val_t route_neuron_config(el_val_t method, el_val_t path, el_val_t body) {
el_val_t key = query_param(path, EL_STR("key"));
return el_str_concat(el_str_concat(EL_STR("{\"key\":\""), key), EL_STR("\",\"value\":\"\"}"));
return 0;
}
el_val_t route_neuron_state_events(el_val_t method, el_val_t path, el_val_t body) {
if (str_eq(method, EL_STR("GET"))) {
el_val_t limit_str = query_param(path, EL_STR("limit"));
el_val_t limit = ({ el_val_t _if_result_8 = 0; if (str_eq(limit_str, EL_STR(""))) { _if_result_8 = (50); } else { _if_result_8 = (str_to_int(limit_str)); } _if_result_8; });
el_val_t offset_str = query_param(path, EL_STR("offset"));
el_val_t offset = ({ el_val_t _if_result_9 = 0; if (str_eq(offset_str, EL_STR(""))) { _if_result_9 = (0); } else { _if_result_9 = (str_to_int(offset_str)); } _if_result_9; });
return engram_scan_nodes_by_type_json(EL_STR("InternalStateEvent"), limit, offset);
}
el_val_t content = json_get_string(body, EL_STR("content"));
if (str_eq(content, EL_STR(""))) {
content = body;
}
el_val_t event_label = json_get_string(content, EL_STR("event"));
el_val_t label = ({ el_val_t _if_result_10 = 0; if (str_eq(event_label, EL_STR(""))) { _if_result_10 = (EL_STR("state-event")); } else { _if_result_10 = (event_label); } _if_result_10; });
el_val_t id = engram_node_full(content, EL_STR("InternalStateEvent"), label, el_from_float(0.3), el_from_float(0.3), el_from_float(1.0), EL_STR("Working"), EL_STR("internal-state"));
return el_str_concat(el_str_concat(EL_STR("{\"ok\":true,\"id\":\""), id), EL_STR("\"}"));
return 0;
}
el_val_t route_neuron_processes(el_val_t method, el_val_t path, el_val_t body) {
return EL_STR("{\"ok\":true,\"processes\":[]}");
return 0;
}
el_val_t route_events_next(el_val_t method, el_val_t path, el_val_t body) {
return EL_STR("{\"ok\":true,\"event\":null}");
return 0;
}
el_val_t route_events_ack(el_val_t method, el_val_t path, el_val_t body) {
return EL_STR("{\"ok\":true}");
return 0;
}
el_val_t route_bm25_search(el_val_t method, el_val_t path, el_val_t body) {
el_val_t q = EL_STR("");
if (str_eq(method, EL_STR("GET"))) {
q = query_param(path, EL_STR("q"));
} else {
q = json_get_string(body, EL_STR("query"));
}
if (str_eq(q, EL_STR(""))) {
return EL_STR("{\"error\":\"query is required\"}");
}
el_val_t limit = query_int(path, EL_STR("limit"), 20);
if (limit == 0) {
limit = json_get_int(body, EL_STR("limit"));
}
if (limit == 0) {
limit = 20;
}
return bm25_search_json(q, limit);
return 0;
}
el_val_t check_auth_ok(el_val_t method, el_val_t body) {
el_val_t key = env(EL_STR("ENGRAM_API_KEY"));
if (str_eq(key, EL_STR(""))) {
@@ -900,60 +299,6 @@ el_val_t handle_request(el_val_t method, el_val_t path, el_val_t body) {
return route_health(method, path, body);
}
}
if (str_starts_with(clean, EL_STR("/api/neuron/")) || str_starts_with(clean, EL_STR("/events/"))) {
if (str_eq(clean, EL_STR("/api/neuron/session/begin"))) {
return route_neuron_session_begin(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/ctx"))) {
return route_neuron_ctx(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/memory"))) {
return route_neuron_memory(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/capture"))) {
return route_neuron_knowledge_capture(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/evolve"))) {
return route_neuron_knowledge_evolve(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/promote"))) {
return route_neuron_knowledge_promote(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/recall"))) {
return route_neuron_recall(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/graph/link"))) {
return route_neuron_graph_link(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/graph"))) {
return route_neuron_graph(method, path, body);
}
if (str_starts_with(clean, EL_STR("/api/neuron/list/"))) {
return route_neuron_list(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/consolidate"))) {
return route_neuron_consolidate(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/config"))) {
return route_neuron_config(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/state-events"))) {
return route_neuron_state_events(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/processes/define"))) {
return route_neuron_processes(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/processes"))) {
return route_neuron_processes(method, path, body);
}
if (str_eq(clean, EL_STR("/events/next"))) {
return route_events_next(method, path, body);
}
if (str_eq(clean, EL_STR("/events/ack"))) {
return route_events_ack(method, path, body);
}
return err_json(EL_STR("not found"));
}
if (!check_auth_ok(method, body)) {
return err_json(EL_STR("unauthorized"));
}
@@ -993,80 +338,15 @@ el_val_t handle_request(el_val_t method, el_val_t path, el_val_t body) {
if (str_eq(method, EL_STR("GET")) && str_starts_with(clean, EL_STR("/api/search"))) {
return route_search(method, path, body);
}
if (str_eq(clean, EL_STR("/api/bm25/search"))) {
return route_bm25_search(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && (str_eq(clean, EL_STR("/api/strengthen")) || str_eq(clean, EL_STR("/strengthen")))) {
return route_strengthen(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && ((str_eq(clean, EL_STR("/api/decay")) || str_eq(clean, EL_STR("/api/maintenance"))) || str_eq(clean, EL_STR("/decay")))) {
return route_decay(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && (str_eq(clean, EL_STR("/api/export")) || str_eq(clean, EL_STR("/export")))) {
return route_export(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && (str_eq(clean, EL_STR("/api/save")) || str_eq(clean, EL_STR("/save")))) {
return route_export(method, path, body);
return route_save(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && (str_eq(clean, EL_STR("/api/load")) || str_eq(clean, EL_STR("/load")))) {
return route_load(method, path, body);
}
if (str_eq(method, EL_STR("POST")) && (str_eq(clean, EL_STR("/api/reindex")) || str_eq(clean, EL_STR("/reindex")))) {
return route_reindex(method, path, body);
}
if (str_starts_with(clean, EL_STR("/api/neuron/"))) {
if (str_eq(clean, EL_STR("/api/neuron/session/begin"))) {
return route_neuron_session_begin(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/ctx"))) {
return route_neuron_ctx(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/memory"))) {
return route_neuron_memory(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/capture"))) {
return route_neuron_knowledge_capture(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/evolve"))) {
return route_neuron_knowledge_evolve(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/knowledge/promote"))) {
return route_neuron_knowledge_promote(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/recall"))) {
return route_neuron_recall(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/graph/link"))) {
return route_neuron_graph_link(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/graph"))) {
return route_neuron_graph(method, path, body);
}
if (str_starts_with(clean, EL_STR("/api/neuron/list/"))) {
return route_neuron_list(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/consolidate"))) {
return route_neuron_consolidate(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/config"))) {
return route_neuron_config(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/state-events"))) {
return route_neuron_state_events(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/processes/define"))) {
return route_neuron_processes(method, path, body);
}
if (str_eq(clean, EL_STR("/api/neuron/processes"))) {
return route_neuron_processes(method, path, body);
}
}
if (str_eq(clean, EL_STR("/events/next"))) {
return route_events_next(method, path, body);
}
if (str_eq(clean, EL_STR("/events/ack"))) {
return route_events_ack(method, path, body);
}
return el_str_concat(el_str_concat(EL_STR("{\"error\":\"not found\",\"path\":\""), clean), EL_STR("\"}"));
return 0;
}
@@ -1082,20 +362,9 @@ int main(int _argc, char** _argv) {
if (str_eq(data_dir, EL_STR(""))) {
data_dir = EL_STR("/tmp/engram");
}
db_path = el_str_concat(data_dir, EL_STR("/engram.db"));
loaded = engram_load_binary_el(db_path);
if (!loaded) {
engram_load_dir(data_dir);
if (engram_node_count() == 0) {
el_val_t snapshot_path = el_str_concat(data_dir, EL_STR("/snapshot.json"));
engram_load(snapshot_path);
}
if (engram_node_count() > 0) {
engram_write_binary_el(db_path);
println(EL_STR("[engram] migrated legacy data to binary format"));
}
}
println(EL_STR("[engram] runtime-native graph engine (ML-KEM-1024 encrypted)"));
snapshot_path = el_str_concat(data_dir, EL_STR("/snapshot.json"));
engram_load(snapshot_path);
println(EL_STR("[engram] runtime-native graph engine"));
println(el_str_concat(EL_STR("[engram] data_dir="), data_dir));
println(el_str_concat(EL_STR("[engram] node_count="), int_to_str(engram_node_count())));
println(el_str_concat(EL_STR("[engram] edge_count="), int_to_str(engram_edge_count())));
+20 -799
View File
@@ -6,297 +6,15 @@
// database.
//
// Built and linked with:
// elc src/server.el > ../dist/engram.c
// cc -std=c11 -O2 \
// -I/Users/will/Development/neuron-technologies/foundation/el/lang/releases/v1.0.0-20260501 \
// -I/opt/homebrew/Cellar/liboqs/0.15.0/include \
// -I/opt/homebrew/opt/openssl@3/include \
// -L/opt/homebrew/Cellar/liboqs/0.15.0/lib \
// -L/opt/homebrew/opt/openssl@3/lib \
// -lcurl -lpthread -loqs -lssl -lcrypto \
// -o ../dist/engram ../dist/engram.c \
// /Users/will/Development/neuron-technologies/foundation/el/lang/releases/v1.0.0-20260501/el_runtime.c
// ./dist/engram
// elc src/server.el > server.c
// cc -std=c11 -O2 -lcurl -lpthread -o engram server.c el_runtime.c
// ./engram
//
// Configuration via environment:
// ENGRAM_BIND host:port (default :8742)
// ENGRAM_API_KEY bearer auth (optional)
// ENGRAM_DATA_DIR snapshot location (default ~/.neuron/engram)
// BM25+ text ranking
//
// Implements BM25+ (Lv & Zhai 2011) for in-process keyword search over the
// engram node store. No external dependencies pure EL, zero Ollama calls.
//
// Parameters: k1=1.2, b=0.75, delta=1.0
//
// V1 simplification: n(t) (number of docs containing term t) is approximated
// as 1 for all terms. This collapses IDF to a constant per corpus size:
// IDF = ln((N - 1 + 0.5) / (1 + 0.5) + 1) = ln((N + 0.5) / 1.5 + 1)
// Scoring effectively becomes TF-length-normalised BM25+ (delta term present).
// Acceptable for V1; a real inverted index can replace this later.
fn bm25_tokenize(text: String) -> String {
// Lowercase and strip punctuation (replace with spaces), then trim.
let t: String = str_to_lower(text)
let t = str_replace(t, ".", " ")
let t = str_replace(t, ",", " ")
let t = str_replace(t, "!", " ")
let t = str_replace(t, "?", " ")
let t = str_replace(t, "\"", " ")
let t = str_replace(t, ":", " ")
let t = str_replace(t, ";", " ")
let t = str_replace(t, "(", " ")
let t = str_replace(t, ")", " ")
let t = str_replace(t, "[", " ")
let t = str_replace(t, "]", " ")
let t = str_replace(t, "{", " ")
let t = str_replace(t, "}", " ")
let t = str_replace(t, "/", " ")
let t = str_replace(t, "\\", " ")
let t = str_replace(t, "'", " ")
let t = str_replace(t, "-", " ")
let t = str_replace(t, "_", " ")
let t = str_replace(t, "+", " ")
str_trim(t)
}
fn bm25_count_term(term: String, doc_tokens: String) -> Int {
// Pad with spaces to avoid prefix/suffix partial matches.
let padded_term: String = " " + term + " "
let padded_doc: String = " " + doc_tokens + " "
str_count(padded_doc, padded_term)
}
fn bm25_score_doc(doc_content: String, query_tokens: String, corpus_size: Int, avg_doc_len: String) -> String {
// BM25+ parameters (stored as strings = float-encoded el_val_t from el_from_float)
// We use float_add/float_mul/float_div builtins to avoid EL operator issues.
// avg_doc_len is passed as a String slot holding an el_val_t float bit-pattern.
// (EL has no safe float-passing convention; we work around using str_to_float.)
//
// V1: n_t=1 for all terms. IDF = ln((N+0.5)/1.5 + 1) = constant per corpus.
// This collapses BM25+ to TF-length-normalised scoring acceptable for V1.
let k1: Float = 1.2
let b: Float = 0.75
let delta: Float = 1.0
let doc_tokens: String = bm25_tokenize(doc_content)
let doc_wc: Int = str_count_words(doc_tokens)
if doc_wc == 0 { return "0.0" }
let doc_len: Float = int_to_float(doc_wc)
let avg_len: Float = str_to_float(avg_doc_len)
// IDF constant
let N: Float = int_to_float(corpus_size)
// (N + 0.5) / 1.5 + 1.0
let idf_arg: Float = float_add(float_div(float_add(N, 1.2), 1.5), 1.0)
let idf: Float = math_log(idf_arg)
// Sum TF component over query terms
let terms: List = str_split(query_tokens, " ")
let n_terms: Int = len(terms)
let score: Float = 0.0
let i: Int = 0
while i < n_terms {
let term: String = get(terms, i)
let tlen: Int = str_len(term)
if tlen >= 2 {
let tf_count: Int = bm25_count_term(term, doc_tokens)
if tf_count > 0 {
let tf_raw: Float = int_to_float(tf_count)
// norm_factor = 1 - b + b * doc_len / avg_len
let norm_factor: Float = float_add(float_sub(1.0, b), float_div(float_mul(b, doc_len), avg_len))
// tf_comp = delta + tf * (k1+1) / (tf + k1*norm)
let numerator: Float = float_mul(tf_raw, float_add(k1, 1.0))
let denominator: Float = float_add(tf_raw, float_mul(k1, norm_factor))
let tf_comp: Float = float_add(delta, float_div(numerator, denominator))
let score = float_add(score, float_mul(idf, tf_comp))
}
}
let i = i + 1
}
// Return score as a string so it survives EL's lack of float-in-list support
float_to_str(score)
}
fn bm25_search_json(query: String, limit: Int) -> String {
// 1. Determine scan size: floor at 200 so small `limit` values still scan
// enough of the corpus to find relevant nodes.
// Cap raised from 500 5000 (2026-05-24 self-review): 500 was 0.3% of the
// 161K-node corpus. At 5000 we cover the top-3% by salience still fast
// (pure C scan, no Ollama calls) and 10x better recall for content search.
// engram_scan_nodes_json returns nodes sorted by salience DESC, so ISEs
// (salience 0.3) naturally fall below Knowledge/Memory (0.50.8), keeping
// the effective search corpus content-dense.
let scan_limit: Int = limit * 10
if scan_limit < 200 { let scan_limit = 200 }
if scan_limit > 5000 { let scan_limit = 5000 }
// 2. Fetch node sample
let nodes_json: String = engram_scan_nodes_json(scan_limit, 0)
let n: Int = json_array_len(nodes_json)
if n == 0 { return "[]" }
// 3. Compute avg_doc_len from sample
let total_words: Int = 0
let i: Int = 0
while i < n {
let node: String = json_array_get(nodes_json, i)
let content: String = json_get_string(node, "content")
let tokens: String = bm25_tokenize(content)
let wc: Int = str_count_words(tokens)
let total_words = total_words + wc
let i = i + 1
}
// avg_doc_len as string for safe float passing
let avg_doc_len_f: Float = float_div(int_to_float(total_words), int_to_float(n))
let avg_doc_len: String = if float_gt(avg_doc_len_f, 0.0) { float_to_str(avg_doc_len_f) } else { "1.0" }
// 4. Tokenize query
let query_tokens: String = bm25_tokenize(query)
if str_eq(str_trim(query_tokens), "") { return "[]" }
// 5. Score each node; collect results as parallel JSON and score lists.
// Scores are stored as strings (float_to_str) to avoid float-in-list issues.
let result_nodes: List = 0
let result_scores: List = 0
let result_count: Int = 0
let j: Int = 0
while j < n {
let node: String = json_array_get(nodes_json, j)
let content: String = json_get_string(node, "content")
let sc_str: String = bm25_score_doc(content, query_tokens, n, avg_doc_len)
// Only include nodes with score > 0.0 (use float comparison, not string match
// float_to_str(0.0) returns "0.000000", not "0.0").
if float_gt(str_to_float(sc_str), 0.0) {
let result_nodes = list_push(result_nodes, node)
let result_scores = list_push(result_scores, sc_str)
let result_count = result_count + 1
}
let j = j + 1
}
if result_count == 0 { return "[]" }
// 6. Selection-sort descending by score, take top `limit`
let out_limit: Int = if result_count < limit { result_count } else { limit }
let k: Int = 0
while k < out_limit {
// Find max score index in [k, result_count)
let max_idx: Int = k
let max_sc_str: String = get(result_scores, k)
let max_sc_f: Float = str_to_float(max_sc_str)
let p: Int = k + 1
while p < result_count {
let sc2_str: String = get(result_scores, p)
let sc2_f: Float = str_to_float(sc2_str)
if float_gt(sc2_f, max_sc_f) {
let max_sc_f = sc2_f
let max_sc_str = sc2_str
let max_idx = p
}
let p = p + 1
}
// Swap k <-> max_idx
if max_idx != k {
let tmp_node: String = get(result_nodes, k)
let tmp_sc: String = get(result_scores, k)
let result_nodes = list_set(result_nodes, k, get(result_nodes, max_idx))
let result_scores = list_set(result_scores, k, get(result_scores, max_idx))
let result_nodes = list_set(result_nodes, max_idx, tmp_node)
let result_scores = list_set(result_scores, max_idx, tmp_sc)
}
let k = k + 1
}
// 7. Build JSON array of top `out_limit` nodes with bm25_score field
let out: String = "["
let r: Int = 0
while r < out_limit {
let node: String = get(result_nodes, r)
let sc_str: String = get(result_scores, r)
// Inject bm25_score: trim the closing } and append field
let node_len: Int = str_len(node)
let node_body: String = str_slice(node, 0, node_len - 1)
let entry: String = node_body + ",\"bm25_score\":" + sc_str + "}"
if r > 0 { let out = out + "," }
let out = out + entry
let r = r + 1
}
out + "]"
}
// Auto-linking
//
// auto_link_content_node link a newly-created Knowledge or Memory node to
// semantically related non-ISE nodes via BM25 search.
//
// Problem it solves: route_neuron_memory and route_neuron_knowledge_capture
// both call engram_node_full directly, creating nodes with zero edges. With
// 14K+ ISEs dominating the corpus, BFS traversal contributes nothing every
// query relies solely on lexical/semantic seed matching. Auto-linking builds
// explicit "related" edges so activated knowledge nodes fan out to connected
// neighbors during BFS.
//
// Design choices:
// - BM25 (not substring search): ranks by relevance, not just occurrence
// - Skip InternalStateEvent nodes: ISEs dominate the corpus and are not
// useful link targets for knowledge/memory nodes
// - Up to 3 edges per node: enough to build graph structure without over-linking
// - weight=0.6: moderately strong; causal edges (field-validated at 2.0) are
// much stronger, so these "related" edges don't flood activation paths
// - state_set for linked counter: EL `let` in nested if-blocks creates inner
// scope only; state_set persists across block boundaries (2026-05-25 lesson)
//
// (2026-05-28 self-review)
fn auto_link_content_node(node_id: String, content: String) -> Int {
let clen: Int = str_len(content)
if clen < 20 { return 0 }
// Find search term: first word >= 5 chars, or second word.
let sp1: Int = str_index_of(content, " ")
let w1end: Int = if sp1 < 0 { clen } else { sp1 }
let word1: String = str_slice(content, 0, w1end)
state_set("aln_term", "")
if str_len(word1) >= 5 {
state_set("aln_term", word1)
}
if str_eq(state_get("aln_term"), "") {
if sp1 >= 0 {
let rest: String = str_slice(content, sp1 + 1, clen)
let sp2: Int = str_index_of(rest, " ")
let w2end: Int = if sp2 < 0 { str_len(rest) } else { sp2 }
let word2: String = str_slice(rest, 0, w2end)
if str_len(word2) >= 5 {
state_set("aln_term", word2)
}
}
}
let search_term: String = state_get("aln_term")
if str_eq(search_term, "") { return 0 }
// BM25 over top-20 results; skip ISE nodes; connect up to 3.
let results: String = bm25_search_json(search_term, 20)
let n: Int = json_array_len(results)
state_set("aln_linked", "0")
let i: Int = 0
while i < n {
let linked_so_far: Int = str_to_int(state_get("aln_linked"))
if linked_so_far < 3 {
let elem: String = json_array_get(results, i)
let rid: String = json_get_string(elem, "id")
let rtype: String = json_get_string(elem, "node_type")
if !str_eq(rtype, "InternalStateEvent") && !str_eq(rid, "") && !str_eq(rid, node_id) {
engram_connect(node_id, rid, 0.6, "related")
state_set("aln_linked", int_to_str(linked_so_far + 1))
}
}
let i = i + 1
}
return str_to_int(state_get("aln_linked"))
}
// Helpers
fn parse_port(bind: String) -> Int {
@@ -365,13 +83,7 @@ fn route_create_node(method: String, path: String, body: String) -> String {
let salience: Float = json_get_float(body, "salience")
if salience == 0.0 { let salience = 0.5 }
let id: String = engram_node(content, node_type, salience)
// Auto-link via BM25 search reuse auto_link_content_node which skips
// ISE nodes and links to up to 3 semantically related non-ISE nodes.
// Replaces the old inline substring-search auto-link (2026-05-29 cleanup).
let auto_linked: Int = auto_link_content_node(id, content)
"{\"id\":\"" + id + "\",\"content\":\"" + content + "\",\"node_type\":\"" + node_type + "\",\"auto_linked\":" + int_to_str(auto_linked) + "}"
"{\"id\":\"" + id + "\",\"content\":\"" + content + "\",\"node_type\":\"" + node_type + "\"}"
}
fn route_get_node(method: String, path: String, body: String) -> String {
@@ -419,7 +131,7 @@ fn route_search(method: String, path: String, body: String) -> String {
let limit: Int = query_int(path, "limit", 20)
if limit == 0 { let limit = json_get_int(body, "limit") }
if limit == 0 { let limit = 20 }
return bm25_search_json(q, limit)
return engram_search_json(q, limit)
}
fn route_activate(method: String, path: String, body: String) -> String {
@@ -433,17 +145,6 @@ fn route_activate(method: String, path: String, body: String) -> String {
let bd: Int = json_get_int(body, "depth")
if bd > 0 { let depth = bd }
}
// BM25 pre-bias: strengthen top-10 BM25 results before spreading activation
// so semantically relevant nodes already have elevated salience.
let top: String = bm25_search_json(q, 10)
let nb: Int = json_array_len(top)
let bi: Int = 0
while bi < nb {
let node: String = json_array_get(top, bi)
let nid: String = json_get_string(node, "id")
if !str_eq(nid, "") { engram_strengthen(nid) }
let bi = bi + 1
}
return "{\"results\":" + engram_activate_json(q, depth) + "}"
}
@@ -479,41 +180,25 @@ fn route_forget(method: String, path: String, body: String) -> String {
ok_json()
}
fn route_decay(method: String, path: String, body: String) -> String {
engram_apply_decay_json()
}
fn route_export(method: String, path: String, body: String) -> String {
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
// Write binary checkpoint
let db_path: String = dir + "/engram.db"
engram_write_binary_el(db_path)
// Also write JSON export for human inspection
fn route_save(method: String, path: String, body: String) -> String {
let p: String = json_get_string(body, "path")
if str_eq(p, "") {
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let p = dir + "/snapshot.json"
}
engram_save(p)
"{\"ok\":true,\"binary\":\"" + db_path + "\",\"json\":\"" + p + "\"}"
}
fn route_reindex(method: String, path: String, body: String) -> String {
engram_reindex_json()
"{\"ok\":true,\"path\":\"" + p + "\"}"
}
fn route_load(method: String, path: String, body: String) -> String {
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let db_path: String = dir + "/engram.db"
let ok: Bool = engram_load_binary_el(db_path)
if !ok {
let p: String = json_get_string(body, "path")
if str_eq(p, "") {
let p = dir + "/snapshot.json"
}
engram_load(p)
let p: String = json_get_string(body, "path")
if str_eq(p, "") {
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let p = dir + "/snapshot.json"
}
engram_load(p)
ok_json()
}
@@ -521,323 +206,6 @@ fn route_health(method: String, path: String, body: String) -> String {
"{\"status\":\"ok\",\"engine\":\"engram-runtime-native\"}"
}
// /api/neuron/* Routes
// route_neuron_session_begin activate with broad seeds, return node stats + results
fn route_neuron_session_begin(method: String, path: String, body: String) -> String {
let results: String = engram_activate_json("memory knowledge context", 2)
let nc: Int = engram_node_count()
let ec: Int = engram_edge_count()
"{\"ok\":true,\"nodes\":" + results + ",\"node_count\":" + int_to_str(nc) + ",\"edge_count\":" + int_to_str(ec) + "}"
}
// route_neuron_ctx compile working context from top activated nodes
fn route_neuron_ctx(method: String, path: String, body: String) -> String {
let results: String = engram_activate_json("architecture decision memory", 2)
let n: Int = json_array_len(results)
let limit: Int = if n > 10 { 10 } else { n }
let ctx: String = "Recent working memory:\n"
let i: Int = 0
let ctx_body: String = ""
while i < limit {
let elem: String = json_array_get(results, i)
let label: String = json_get_string(elem, "label")
let content: String = json_get_string(elem, "content")
let clen: Int = str_len(content)
let snippet: String = if clen > 200 { str_slice(content, 0, 200) } else { content }
let ctx_body = ctx_body + "- [" + label + "]: " + snippet + "\n"
let i = i + 1
}
let full_ctx: String = ctx + ctx_body
"{\"ok\":true,\"context\":\"" + str_replace(str_replace(str_replace(full_ctx, "\\", "\\\\"), "\"", "\\\""), "\n", "\\n") + "\"}"
}
// route_neuron_memory create a Memory node with importance-to-tier mapping
fn route_neuron_memory(method: String, path: String, body: String) -> String {
let content: String = json_get_string(body, "content")
if str_eq(content, "") { return "{\"error\":\"content is required\"}" }
let node_type: String = json_get_string(body, "node_type")
if str_eq(node_type, "") { let node_type = "Memory" }
let label: String = json_get_string(body, "label")
let importance: String = json_get_string(body, "importance")
let project: String = json_get_string(body, "project")
let tags_raw: String = json_get_string(body, "tags")
// Map importance to tier
let tier: String = "Episodic"
if str_eq(importance, "critical") { let tier = "Procedural" }
if str_eq(importance, "high") { let tier = "Semantic" }
if str_eq(importance, "normal") { let tier = "Episodic" }
if str_eq(importance, "low") { let tier = "Working" }
// Override with explicit tier if provided
let explicit_tier: String = json_get_string(body, "tier")
if !str_eq(explicit_tier, "") { let tier = explicit_tier }
// Build tags string append project tag if set
let tags_str: String = tags_raw
if !str_eq(project, "") {
if str_eq(tags_str, "") {
let tags_str = "project:" + project
}
if !str_eq(tags_str, "") {
let tags_str = tags_str + " project:" + project
}
}
let id: String = engram_node_full(content, node_type, label, 0.5, 0.5, 1.0, tier, tags_str)
// Auto-link to related non-ISE nodes so this memory is reachable via BFS traversal.
// Without this, MCP-created nodes arrive with zero edges and are invisible to
// graph spread during activation (only lexical/semantic seed matching finds them).
let auto_linked: Int = auto_link_content_node(id, content)
// Checkpoint after write
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let db_path: String = dir + "/engram.db"
engram_write_binary_el(db_path)
"{\"ok\":true,\"id\":\"" + id + "\",\"auto_linked\":" + int_to_str(auto_linked) + ",\"content\":\"" + str_replace(str_replace(content, "\\", "\\\\"), "\"", "\\\"") + "\"}"
}
// route_neuron_knowledge_capture create a Knowledge node
fn route_neuron_knowledge_capture(method: String, path: String, body: String) -> String {
let content: String = json_get_string(body, "content")
if str_eq(content, "") { return "{\"error\":\"content is required\"}" }
let title: String = json_get_string(body, "title")
let category: String = json_get_string(body, "category")
let tags_raw: String = json_get_string(body, "tags")
let project: String = json_get_string(body, "project")
let tier_raw: String = json_get_string(body, "tier")
// Map tier name to engram tier
let tier: String = "Episodic"
if str_eq(tier_raw, "lesson") { let tier = "Semantic" }
if str_eq(tier_raw, "canonical") { let tier = "Procedural" }
if str_eq(tier_raw, "note") { let tier = "Episodic" }
// Build tags
let tags_str: String = tags_raw
if !str_eq(category, "") {
if str_eq(tags_str, "") {
let tags_str = "category:" + category
}
if !str_eq(tags_str, "") {
let tags_str = tags_str + " category:" + category
}
}
if !str_eq(project, "") {
if str_eq(tags_str, "") {
let tags_str = "project:" + project
}
if !str_eq(tags_str, "") {
let tags_str = tags_str + " project:" + project
}
}
let id: String = engram_node_full(content, "Knowledge", title, 0.7, 0.7, 1.0, tier, tags_str)
// Auto-link to related non-ISE nodes for BFS reachability (same rationale as route_neuron_memory).
let auto_linked: Int = auto_link_content_node(id, content)
// Checkpoint
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let db_path: String = dir + "/engram.db"
engram_write_binary_el(db_path)
"{\"ok\":true,\"id\":\"" + id + "\",\"auto_linked\":" + int_to_str(auto_linked) + "}"
}
// route_neuron_knowledge_evolve create updated node (evolution via new node)
fn route_neuron_knowledge_evolve(method: String, path: String, body: String) -> String {
let content: String = json_get_string(body, "content")
let prior_id: String = json_get_string(body, "id")
if str_eq(content, "") { return "{\"ok\":true}" }
let id: String = engram_node_full(content, "Knowledge", "", 0.7, 0.7, 1.0, "Semantic", "evolved")
if !str_eq(prior_id, "") && !str_eq(id, "") {
engram_connect(id, prior_id, 1.0, "supersedes")
}
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
engram_write_binary_el(dir + "/engram.db")
"{\"ok\":true,\"id\":\"" + id + "\"}"
}
// route_neuron_knowledge_promote promote a knowledge node to a higher tier.
// Creates a new node with the promoted tier (same content) and connects
// via a "supersedes" edge from new old. Tier mapping:
// note/Episodic lesson/Semantic canonical/Procedural
fn route_neuron_knowledge_promote(method: String, path: String, body: String) -> String {
let id: String = json_get_string(body, "id")
if str_eq(id, "") { return "{\"ok\":true}" }
// Read existing node
let node_json: String = engram_get_node_json(id)
if str_eq(node_json, "") { return err_json("node not found") }
if str_eq(node_json, "null") { return err_json("node not found") }
let content: String = json_get_string(node_json, "content")
if str_eq(content, "") { return err_json("node has no content") }
let label: String = json_get_string(node_json, "label")
let tags: String = json_get_string(node_json, "tags")
let current_tier: String = json_get_string(node_json, "tier")
// Determine target tier: explicit override or auto-promote one level
let tier_raw: String = json_get_string(body, "tier")
let new_tier: String = ""
// Explicit tier takes precedence
if str_eq(tier_raw, "lesson") { let new_tier = "Semantic" }
if str_eq(tier_raw, "canonical") { let new_tier = "Procedural" }
if str_eq(tier_raw, "note") { let new_tier = "Episodic" }
// Auto-promote one level if no explicit tier
if str_eq(new_tier, "") {
if str_eq(current_tier, "Working") { let new_tier = "Episodic" }
if str_eq(current_tier, "Episodic") { let new_tier = "Semantic" }
if str_eq(current_tier, "Semantic") { let new_tier = "Procedural" }
if str_eq(current_tier, "Procedural") { let new_tier = "Procedural" }
}
if str_eq(new_tier, "") { let new_tier = "Semantic" }
// Create promoted node higher importance (0.8) signals durable knowledge
let new_id: String = engram_node_full(content, "Knowledge", label, 0.7, 0.8, 1.0, new_tier, tags)
// Wire supersedes edge: new node supersedes old
if !str_eq(new_id, "") {
engram_connect(new_id, id, 1.0, "supersedes")
}
// Checkpoint
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
engram_write_binary_el(dir + "/engram.db")
"{\"ok\":true,\"id\":\"" + new_id + "\",\"promoted_from\":\"" + id + "\",\"tier\":\"" + new_tier + "\"}"
}
// route_neuron_recall search or list nodes
fn route_neuron_recall(method: String, path: String, body: String) -> String {
let query: String = json_get_string(body, "query")
let chain: String = json_get_string(body, "chain_name")
let limit: Int = json_get_int(body, "limit")
if limit == 0 { let limit = 20 }
let q: String = if str_eq(query, "") { chain } else { query }
if str_eq(q, "") {
return engram_scan_nodes_json(limit, 0)
}
return bm25_search_json(q, limit)
}
// route_neuron_graph get node + search-based neighbor approximation.
// engram_neighbors_json crashes on large graphs (15k+ edges exceeds BFS cap).
// Use a search-based approach instead: search by the node id string, which
// returns connected nodes that share content with the target id in edges/tags.
// For the mcp-wrapper callers this is sufficient they just need the node itself.
fn route_neuron_graph(method: String, path: String, body: String) -> String {
let id: String = query_param(path, "id")
if str_eq(id, "") { return "{\"error\":\"id is required\"}" }
let node_json: String = engram_get_node_json(id)
// Return node with empty neighbors safe fallback avoids BFS crash
"{\"ok\":true,\"node\":" + node_json + ",\"neighbors\":[]}"
}
// route_neuron_graph_link create edge between nodes
fn route_neuron_graph_link(method: String, path: String, body: String) -> String {
let from_id: String = json_get_string(body, "from_id")
let to_id: String = json_get_string(body, "to_id")
if str_eq(from_id, "") || str_eq(to_id, "") {
return "{\"error\":\"from_id and to_id are required\"}"
}
let relation: String = json_get_string(body, "relation")
if str_eq(relation, "") { let relation = "related" }
let weight: Float = json_get_float(body, "weight")
if weight == 0.0 { let weight = 0.5 }
engram_connect(from_id, to_id, weight, relation)
"{\"ok\":true,\"from_id\":\"" + from_id + "\",\"to_id\":\"" + to_id + "\",\"relation\":\"" + relation + "\"}"
}
// route_neuron_list list nodes by type extracted from path
fn route_neuron_list(method: String, path: String, body: String) -> String {
let clean: String = strip_query(path)
let prefix: String = "/api/neuron/list/"
let node_type: String = str_slice(clean, str_len(prefix), str_len(clean))
let limit: Int = query_int(path, "limit", 50)
if str_eq(node_type, "") { return "[]" }
return engram_scan_nodes_by_type_json(node_type, limit, 0)
}
// route_neuron_consolidate checkpoint and return counts
fn route_neuron_consolidate(method: String, path: String, body: String) -> String {
let dir: String = env("ENGRAM_DATA_DIR")
if str_eq(dir, "") { let dir = "/tmp/engram" }
let db_path: String = dir + "/engram.db"
engram_write_binary_el(db_path)
let nc: Int = engram_node_count()
let ec: Int = engram_edge_count()
"{\"ok\":true,\"node_count\":" + int_to_str(nc) + ",\"edge_count\":" + int_to_str(ec) + "}"
}
// route_neuron_config return stub config values
fn route_neuron_config(method: String, path: String, body: String) -> String {
let key: String = query_param(path, "key")
"{\"key\":\"" + key + "\",\"value\":\"\"}"
}
// route_neuron_state_events GET lists ISEs, POST logs a new one.
// GET supports ?limit=N&offset=M for pagination; ?label=X to extract label
// from the ISE content's "event" field.
// ISEs sort by created_at DESC (most-recent-first) as of 2026-05-23 fix.
// ?limit=10 returns the 10 most recent ISEs. Offset for pagination, not for
// skipping to recent events (that was the pre-fix behavior; no longer needed).
fn route_neuron_state_events(method: String, path: String, body: String) -> String {
if str_eq(method, "GET") {
let limit_str: String = query_param(path, "limit")
let limit: Int = if str_eq(limit_str, "") { 50 } else { str_to_int(limit_str) }
let offset_str: String = query_param(path, "offset")
let offset: Int = if str_eq(offset_str, "") { 0 } else { str_to_int(offset_str) }
return engram_scan_nodes_by_type_json("InternalStateEvent", limit, offset)
}
let content: String = json_get_string(body, "content")
if str_eq(content, "") { let content = body }
// Extract label from content JSON "event" field for better ISE searchability
let event_label: String = json_get_string(content, "event")
let label: String = if str_eq(event_label, "") { "state-event" } else { event_label }
let id: String = engram_node_full(content, "InternalStateEvent", label, 0.3, 0.3, 1.0, "Working", "internal-state")
"{\"ok\":true,\"id\":\"" + id + "\"}"
}
// route_neuron_processes stub
fn route_neuron_processes(method: String, path: String, body: String) -> String {
"{\"ok\":true,\"processes\":[]}"
}
// route_events_next stub empty event queue
fn route_events_next(method: String, path: String, body: String) -> String {
"{\"ok\":true,\"event\":null}"
}
// route_events_ack stub ack
fn route_events_ack(method: String, path: String, body: String) -> String {
"{\"ok\":true}"
}
fn route_bm25_search(method: String, path: String, body: String) -> String {
let q: String = ""
if str_eq(method, "GET") {
let q = query_param(path, "q")
} else {
let q = json_get_string(body, "query")
}
if str_eq(q, "") { return "{\"error\":\"query is required\"}" }
let limit: Int = query_int(path, "limit", 20)
if limit == 0 { let limit = json_get_int(body, "limit") }
if limit == 0 { let limit = 20 }
bm25_search_json(q, limit)
}
// Auth
fn check_auth_ok(method: String, body: String) -> Bool {
@@ -864,63 +232,6 @@ fn handle_request(method: String, path: String, body: String) -> String {
}
}
// /api/neuron/* and /events/* are pre-auth the mcp-wrapper is a trusted
// local service that cannot inject _auth into its request bodies.
if str_starts_with(clean, "/api/neuron/") || str_starts_with(clean, "/events/") {
if str_eq(clean, "/api/neuron/session/begin") {
return route_neuron_session_begin(method, path, body)
}
if str_eq(clean, "/api/neuron/ctx") {
return route_neuron_ctx(method, path, body)
}
if str_eq(clean, "/api/neuron/memory") {
return route_neuron_memory(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/capture") {
return route_neuron_knowledge_capture(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/evolve") {
return route_neuron_knowledge_evolve(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/promote") {
return route_neuron_knowledge_promote(method, path, body)
}
if str_eq(clean, "/api/neuron/recall") {
return route_neuron_recall(method, path, body)
}
if str_eq(clean, "/api/neuron/graph/link") {
return route_neuron_graph_link(method, path, body)
}
if str_eq(clean, "/api/neuron/graph") {
return route_neuron_graph(method, path, body)
}
if str_starts_with(clean, "/api/neuron/list/") {
return route_neuron_list(method, path, body)
}
if str_eq(clean, "/api/neuron/consolidate") {
return route_neuron_consolidate(method, path, body)
}
if str_eq(clean, "/api/neuron/config") {
return route_neuron_config(method, path, body)
}
if str_eq(clean, "/api/neuron/state-events") {
return route_neuron_state_events(method, path, body)
}
if str_eq(clean, "/api/neuron/processes/define") {
return route_neuron_processes(method, path, body)
}
if str_eq(clean, "/api/neuron/processes") {
return route_neuron_processes(method, path, body)
}
if str_eq(clean, "/events/next") {
return route_events_next(method, path, body)
}
if str_eq(clean, "/events/ack") {
return route_events_ack(method, path, body)
}
return err_json("not found")
}
// Auth (when ENGRAM_API_KEY is set)
if !check_auth_ok(method, body) {
return err_json("unauthorized")
@@ -970,93 +281,18 @@ fn handle_request(method: String, path: String, body: String) -> String {
return route_search(method, path, body)
}
// BM25+ text ranking
if str_eq(clean, "/api/bm25/search") {
return route_bm25_search(method, path, body)
}
// Strengthen
if str_eq(method, "POST") && (str_eq(clean, "/api/strengthen") || str_eq(clean, "/strengthen")) {
return route_strengthen(method, path, body)
}
// Temporal decay maintenance
if str_eq(method, "POST") && (str_eq(clean, "/api/decay") || str_eq(clean, "/api/maintenance") || str_eq(clean, "/decay")) {
return route_decay(method, path, body)
}
// Persistence
if str_eq(method, "POST") && (str_eq(clean, "/api/export") || str_eq(clean, "/export")) {
return route_export(method, path, body)
}
// /api/save is kept as a backward-compat alias for /api/export
if str_eq(method, "POST") && (str_eq(clean, "/api/save") || str_eq(clean, "/save")) {
return route_export(method, path, body)
return route_save(method, path, body)
}
if str_eq(method, "POST") && (str_eq(clean, "/api/load") || str_eq(clean, "/load")) {
return route_load(method, path, body)
}
if str_eq(method, "POST") && (str_eq(clean, "/api/reindex") || str_eq(clean, "/reindex")) {
return route_reindex(method, path, body)
}
// /api/neuron/*
if str_starts_with(clean, "/api/neuron/") {
// Specific sub-paths first (longer matches before shorter)
if str_eq(clean, "/api/neuron/session/begin") {
return route_neuron_session_begin(method, path, body)
}
if str_eq(clean, "/api/neuron/ctx") {
return route_neuron_ctx(method, path, body)
}
if str_eq(clean, "/api/neuron/memory") {
return route_neuron_memory(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/capture") {
return route_neuron_knowledge_capture(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/evolve") {
return route_neuron_knowledge_evolve(method, path, body)
}
if str_eq(clean, "/api/neuron/knowledge/promote") {
return route_neuron_knowledge_promote(method, path, body)
}
if str_eq(clean, "/api/neuron/recall") {
return route_neuron_recall(method, path, body)
}
if str_eq(clean, "/api/neuron/graph/link") {
return route_neuron_graph_link(method, path, body)
}
if str_eq(clean, "/api/neuron/graph") {
return route_neuron_graph(method, path, body)
}
if str_starts_with(clean, "/api/neuron/list/") {
return route_neuron_list(method, path, body)
}
if str_eq(clean, "/api/neuron/consolidate") {
return route_neuron_consolidate(method, path, body)
}
if str_eq(clean, "/api/neuron/config") {
return route_neuron_config(method, path, body)
}
if str_eq(clean, "/api/neuron/state-events") {
return route_neuron_state_events(method, path, body)
}
if str_eq(clean, "/api/neuron/processes/define") {
return route_neuron_processes(method, path, body)
}
if str_eq(clean, "/api/neuron/processes") {
return route_neuron_processes(method, path, body)
}
}
// /events/*
if str_eq(clean, "/events/next") {
return route_events_next(method, path, body)
}
if str_eq(clean, "/events/ack") {
return route_events_ack(method, path, body)
}
"{\"error\":\"not found\",\"path\":\"" + clean + "\"}"
}
@@ -1067,28 +303,13 @@ let bind_str: String = env("ENGRAM_BIND")
if str_eq(bind_str, "") { let bind_str = ":8742" }
let port: Int = parse_port(bind_str)
// On startup, load from binary database (ML-KEM-1024 encrypted).
// Falls back to per-file JSON, then snapshot.json for migration from older formats.
// On startup, try to load any existing snapshot (best effort).
let data_dir: String = env("ENGRAM_DATA_DIR")
if str_eq(data_dir, "") { let data_dir = "/tmp/engram" }
let db_path: String = data_dir + "/engram.db"
let loaded: Bool = engram_load_binary_el(db_path)
if !loaded {
// Migration path: try per-file JSON
engram_load_dir(data_dir)
if engram_node_count() == 0 {
// Final fallback: legacy snapshot.json
let snapshot_path: String = data_dir + "/snapshot.json"
engram_load(snapshot_path)
}
// If we loaded anything from legacy format, save as binary immediately
if engram_node_count() > 0 {
engram_write_binary_el(db_path)
println("[engram] migrated legacy data to binary format")
}
}
let snapshot_path: String = data_dir + "/snapshot.json"
engram_load(snapshot_path)
println("[engram] runtime-native graph engine (ML-KEM-1024 encrypted)")
println("[engram] runtime-native graph engine")
println("[engram] data_dir=" + data_dir)
println("[engram] node_count=" + int_to_str(engram_node_count()))
println("[engram] edge_count=" + int_to_str(engram_edge_count()))
BIN
View File
Binary file not shown.
BIN
View File
Binary file not shown.
-12
View File
@@ -4342,9 +4342,6 @@ el_val_t builtin_arity(el_val_t name) {
if (str_eq(name, EL_STR("http_serve"))) {
return 2;
}
if (str_eq(name, EL_STR("http_serve_async"))) {
return 2;
}
if (str_eq(name, EL_STR("http_set_handler"))) {
return 1;
}
@@ -4549,12 +4546,6 @@ el_val_t builtin_arity(el_val_t name) {
if (str_eq(name, EL_STR("engram_load"))) {
return 1;
}
if (str_eq(name, EL_STR("engram_load_dir"))) {
return 1;
}
if (str_eq(name, EL_STR("engram_reindex_json"))) {
return 0;
}
if (str_eq(name, EL_STR("engram_get_node_json"))) {
return 1;
}
@@ -4573,9 +4564,6 @@ el_val_t builtin_arity(el_val_t name) {
if (str_eq(name, EL_STR("engram_stats_json"))) {
return 0;
}
if (str_eq(name, EL_STR("engram_apply_decay_json"))) {
return 0;
}
if (str_eq(name, EL_STR("llm_call"))) {
return 2;
}
+361 -232
View File
@@ -1882,6 +1882,83 @@ el_val_t http_serve_v2(el_val_t port, el_val_t handler) {
return 0;
}
/* ── http_serve_async — non-blocking HTTP server ─────────────────────────── */
/* Runs the accept loop in a background pthread, returns immediately so the
* calling EL script can continue (e.g. to run an awareness loop).
*
* El signature: http_serve_async(port, handler) -> Void */
typedef struct { int sock; } HttpServeAsyncArg;
static void* _http_serve_async_loop(void* raw) {
HttpServeAsyncArg* a = (HttpServeAsyncArg*)raw;
int sock = a->sock;
free(a);
while (1) {
struct sockaddr_in6 cli;
socklen_t clen = sizeof(cli);
int cfd = accept(sock, (struct sockaddr*)&cli, &clen);
if (cfd < 0) {
if (errno == EINTR) continue;
perror("accept"); break;
}
pthread_mutex_lock(&_http_conn_mu);
while (_http_conn_active >= HTTP_MAX_CONNS) {
pthread_cond_wait(&_http_conn_cv, &_http_conn_mu);
}
_http_conn_active++;
pthread_mutex_unlock(&_http_conn_mu);
HttpWorkerArg* arg = malloc(sizeof(HttpWorkerArg));
if (!arg) { close(cfd); continue; }
arg->fd = cfd;
pthread_t tid;
if (pthread_create(&tid, NULL, http_worker, arg) != 0) {
close(cfd); free(arg);
pthread_mutex_lock(&_http_conn_mu);
_http_conn_active--;
pthread_cond_signal(&_http_conn_cv);
pthread_mutex_unlock(&_http_conn_mu);
continue;
}
pthread_detach(tid);
}
close(sock);
return NULL;
}
void http_serve_async(el_val_t port, el_val_t handler) {
const char* hname = EL_CSTR(handler);
if (hname && looks_like_string(handler)) {
http_set_handler(handler);
}
int p = (int)port;
if (p <= 0 || p > 65535) { fprintf(stderr, "http_serve_async: invalid port %d\n", p); return; }
int sock = socket(AF_INET6, SOCK_STREAM, 0);
if (sock < 0) { perror("socket"); return; }
int yes = 1; int no = 0;
setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(yes));
setsockopt(sock, IPPROTO_IPV6, IPV6_V6ONLY, &no, sizeof(no));
struct sockaddr_in6 addr;
memset(&addr, 0, sizeof(addr));
addr.sin6_family = AF_INET6;
addr.sin6_addr = in6addr_any;
addr.sin6_port = htons((uint16_t)p);
if (bind(sock, (struct sockaddr*)&addr, sizeof(addr)) < 0) {
perror("bind"); close(sock); return;
}
if (listen(sock, 64) < 0) { perror("listen"); close(sock); return; }
fprintf(stderr, "[http] async listening on [::]:%d (dual-stack)\n", p);
HttpServeAsyncArg* a = malloc(sizeof(HttpServeAsyncArg));
if (!a) { close(sock); return; }
a->sock = sock;
pthread_t tid;
if (pthread_create(&tid, NULL, _http_serve_async_loop, a) != 0) {
perror("pthread_create"); free(a); close(sock); return;
}
pthread_detach(tid);
/* Returns immediately — caller can now run awareness_run() or any loop. */
}
/* Build the response envelope a 4-arg handler can return. We hand-write
* the JSON so the discriminator key always lands first the runtime's
* http_parse_envelope() detects it via prefix match. headers_json must be
@@ -3173,23 +3250,49 @@ static void jb_puts(JsonBuf* b, const char* s) {
static void jb_emit_escaped(JsonBuf* b, const char* s) {
jb_putc(b, '"');
for (; *s; s++) {
unsigned char c = (unsigned char)*s;
const unsigned char* p = (const unsigned char*)s;
while (*p) {
unsigned char c = *p;
switch (c) {
case '"': jb_puts(b, "\\\""); break;
case '\\': jb_puts(b, "\\\\"); break;
case '\b': jb_puts(b, "\\b"); break;
case '\f': jb_puts(b, "\\f"); break;
case '\n': jb_puts(b, "\\n"); break;
case '\r': jb_puts(b, "\\r"); break;
case '\t': jb_puts(b, "\\t"); break;
case '"': jb_puts(b, "\\\""); p++; break;
case '\\': jb_puts(b, "\\\\"); p++; break;
case '\b': jb_puts(b, "\\b"); p++; break;
case '\f': jb_puts(b, "\\f"); p++; break;
case '\n': jb_puts(b, "\\n"); p++; break;
case '\r': jb_puts(b, "\\r"); p++; break;
case '\t': jb_puts(b, "\\t"); p++; break;
default:
if (c < 0x20) {
char tmp[8];
snprintf(tmp, sizeof(tmp), "\\u%04x", c);
jb_puts(b, tmp);
} else {
p++;
} else if (c < 0x80) {
jb_putc(b, (char)c);
p++;
} else {
/* Multi-byte UTF-8: validate sequence, pass through if valid,
* escape as \u00xx if the start byte is invalid/orphaned. */
int seq_len = 0;
if ((c & 0xE0) == 0xC0) seq_len = 2;
else if ((c & 0xF0) == 0xE0) seq_len = 3;
else if ((c & 0xF8) == 0xF0) seq_len = 4;
if (seq_len >= 2) {
int valid = 1;
for (int i = 1; i < seq_len; i++) {
if ((p[i] & 0xC0) != 0x80) { valid = 0; break; }
}
if (valid) {
for (int i = 0; i < seq_len; i++) jb_putc(b, (char)p[i]);
p += seq_len;
break;
}
}
/* Invalid start byte or truncated sequence — escape it */
char tmp[8];
snprintf(tmp, sizeof(tmp), "\\u%04x", c);
jb_puts(b, tmp);
p++;
}
break;
}
@@ -7021,83 +7124,6 @@ el_val_t engram_activate(el_val_t query, el_val_t depth) {
g->nodes[i].background_activation = reached[i] ? best_bg[i] : 0.0;
}
/* ── TRAVERSAL INFERENCE: infer A→C edges when A→B→C was traversed ──
* For each pair of edges (AB, BC) where all three nodes were reached,
* create an inferred AC edge with weight = w(AB) * w(BC) * 0.8
* if no AC edge already exists. Cap at 64 new edges per call.
*
* IMPORTANT: collect candidates FIRST into a flat array (no pointers into
* g->edges held across the apply pass), then apply after this avoids
* dangling pointer bugs if engram_grow_edges() reallocs the array. */
{
const int64_t INFER_CAP = 64;
typedef struct { char from[64]; char to[64]; double weight; } InferCandidate;
InferCandidate* cands = malloc((size_t)INFER_CAP * sizeof(InferCandidate));
int64_t ncands = 0;
int64_t snap_ec = g->edge_count;
if (cands) {
for (int64_t e1 = 0; e1 < snap_ec && ncands < INFER_CAP; e1++) {
EngramEdge* ea = &g->edges[e1];
if (!ea->from_id || !ea->to_id) continue;
int64_t ai = engram_find_node_index(ea->from_id);
int64_t bi = engram_find_node_index(ea->to_id);
if (ai < 0 || bi < 0) continue;
if (!reached[ai] || !reached[bi]) continue;
for (int64_t e2 = 0; e2 < snap_ec && ncands < INFER_CAP; e2++) {
if (e2 == e1) continue;
EngramEdge* eb = &g->edges[e2];
if (!eb->from_id || !eb->to_id) continue;
if (strcmp(eb->from_id, ea->to_id) != 0) continue;
int64_t ci = engram_find_node_index(eb->to_id);
if (ci < 0 || !reached[ci]) continue;
if (ai == ci) continue;
int already = 0;
for (int64_t ex = 0; ex < snap_ec; ex++) {
EngramEdge* ee = &g->edges[ex];
if (ee->from_id && ee->to_id &&
strcmp(ee->from_id, ea->from_id) == 0 &&
strcmp(ee->to_id, eb->to_id) == 0) {
already = 1; break;
}
}
if (already) continue;
int dup = 0;
for (int64_t k = 0; k < ncands; k++) {
if (strcmp(cands[k].from, ea->from_id) == 0 &&
strcmp(cands[k].to, eb->to_id) == 0) { dup = 1; break; }
}
if (dup) continue;
double inf_w = ea->weight * eb->weight * 0.8;
if (inf_w < 0.05) inf_w = 0.05;
if (inf_w > 1.0) inf_w = 1.0;
strncpy(cands[ncands].from, ea->from_id, 63); cands[ncands].from[63] = '\0';
strncpy(cands[ncands].to, eb->to_id, 63); cands[ncands].to[63] = '\0';
cands[ncands].weight = inf_w;
ncands++;
}
}
for (int64_t k = 0; k < ncands; k++) {
engram_grow_edges();
EngramEdge* ne = &g->edges[g->edge_count];
memset(ne, 0, sizeof(*ne));
ne->id = engram_new_id();
/* Use strdup (not el_strdup) so these persist beyond the request. */
ne->from_id = strdup(cands[k].from);
ne->to_id = strdup(cands[k].to);
ne->relation = strdup("inferred");
ne->metadata = strdup("{}");
ne->weight = cands[k].weight;
ne->confidence = 0.8;
ne->created_at = now_ms;
ne->updated_at = now_ms;
ne->last_fired = now_ms;
ne->layer_id = ENGRAM_LAYER_DEFAULT;
g->edge_count++;
}
free(cands);
}
}
/* ── PASS 2: executive filter → working memory promotion ──────────── */
/* Step A: collect inhibitory suppressions from fired inhibitory edges.
* Layered consciousness: inhibition is ONLY recorded against targets
@@ -7193,66 +7219,6 @@ el_val_t engram_activate(el_val_t query, el_val_t depth) {
g->nodes[i].working_memory_weight = wm_weights[i];
}
/* ── HEBBIAN STRENGTHENING: fire together, wire together ─────────────
* For each pair of co-promoted nodes (working_memory_weight > 0) that
* share an edge, boost that edge's weight by 0.05 (capped at 1.0).
* Also increment activation_count and update last_activated on promoted
* nodes this is what drives tier migration below. */
for (int64_t i = 0; i < g->node_count; i++) {
if (wm_weights[i] <= 0.0) continue;
EngramNode* n = &g->nodes[i];
n->activation_count++;
n->last_activated = now_ms;
n->updated_at = now_ms;
}
for (int64_t ei = 0; ei < g->edge_count; ei++) {
EngramEdge* e = &g->edges[ei];
if (!e->from_id || !e->to_id) continue;
int64_t src = engram_find_node_index(e->from_id);
int64_t tgt = engram_find_node_index(e->to_id);
if (src < 0 || tgt < 0) continue;
if (wm_weights[src] > 0.0 && wm_weights[tgt] > 0.0) {
e->weight += 0.05;
if (e->weight > 1.0) e->weight = 1.0;
e->last_fired = now_ms;
e->updated_at = now_ms;
}
}
/* ── TIER MIGRATION: promote nodes based on activation_count thresholds ─
* 04 Working
* 519 Episodic
* 2049 Semantic
* 50+ Procedural
* Only upgrade (never downgrade) to preserve earned tier. */
for (int64_t i = 0; i < g->node_count; i++) {
EngramNode* n = &g->nodes[i];
const char* target_tier = NULL;
int64_t ac = n->activation_count;
if (ac >= 50) target_tier = "Procedural";
else if (ac >= 20) target_tier = "Semantic";
else if (ac >= 5) target_tier = "Episodic";
else target_tier = "Working";
if (target_tier && n->tier && strcmp(n->tier, target_tier) != 0) {
/* Only upgrade (Working < Episodic < Semantic < Procedural). */
int cur_rank = 0, new_rank = 0;
if (strcmp(n->tier, "Working") == 0) cur_rank = 0;
else if (strcmp(n->tier, "Episodic") == 0) cur_rank = 1;
else if (strcmp(n->tier, "Semantic") == 0) cur_rank = 2;
else if (strcmp(n->tier, "Procedural") == 0) cur_rank = 3;
if (strcmp(target_tier, "Working") == 0) new_rank = 0;
else if (strcmp(target_tier, "Episodic") == 0) new_rank = 1;
else if (strcmp(target_tier, "Semantic") == 0) new_rank = 2;
else if (strcmp(target_tier, "Procedural") == 0) new_rank = 3;
if (new_rank > cur_rank) {
free(n->tier);
/* Use strdup (not el_strdup) so tier string persists beyond the request. */
n->tier = strdup(target_tier);
n->updated_at = now_ms;
}
}
}
/* ── Collect all background-activated nodes for the return value ────
* Callers see both layers. Context compilation uses only promoted nodes
* (working_memory_weight > 0). Sort: promoted first by wm_weight desc,
@@ -8026,95 +7992,255 @@ el_val_t engram_query_range(el_val_t start_ms_v, el_val_t end_ms_v) {
return el_wrap_str(b.buf);
}
/* ── engram_apply_decay_json — temporal decay maintenance ────────────────────
/* engram_load_merge — like engram_load but WITHOUT resetting the store.
* Reads a JSON snapshot from `path` and adds any nodes/edges not already
* present in the in-memory graph. Dedup is by node id (for nodes) and by
* (from_id, to_id, relation) tuple (for edges).
*
* Iterates ALL nodes and applies temporal decay to their stored `salience`
* field based on time elapsed since `last_activated`:
* Returns (as an EL int) the count of new nodes added. Embeddings are
* intentionally skipped on merged nodes to avoid Ollama delays at runtime;
* auto_link_semantic will handle them when nodes are next activated.
*
* new_salience = current_salience * decay_rate ^ hours_since_activation
*
* where decay_rate defaults to 0.5^(1/168) per hour (half-life one week),
* or the node's own `temporal_decay_rate` if non-zero.
*
* Nodes with temporal_decay_rate == 0 are NOT immune the global default
* applies. To make a node truly immune, set temporal_decay_rate to a very
* small positive value (e.g. 0.0001). Nodes that are "pinned" can be
* identified by a tier of "Procedural" those are skipped.
*
* After updating salience, nodes with salience < 0.05 AND tier == "Working"
* are pruned (deleted) unless they have no content (guard against garbage).
*
* Returns a JSON summary: {"updated": N, "pruned": N} */
el_val_t engram_apply_decay_json(void) {
* Does not merge layers the in-process layer registry is authoritative. */
el_val_t engram_load_merge(el_val_t path) {
const char* p = EL_CSTR(path);
if (!p || !*p) return 0;
FILE* f = fopen(p, "rb");
if (!f) return 0;
fseek(f, 0, SEEK_END);
long sz = ftell(f);
rewind(f);
if (sz <= 0) { fclose(f); return 0; }
char* data = malloc((size_t)sz + 1);
if (!data) { fclose(f); return 0; }
size_t got = fread(data, 1, (size_t)sz, f);
fclose(f);
data[got] = '\0';
EngramStore* g = engram_get();
int64_t now_ms = engram_now_ms();
int64_t updated = 0, pruned = 0;
int64_t added_nodes = 0;
for (int64_t i = 0; i < g->node_count; i++) {
EngramNode* n = &g->nodes[i];
/* Skip Procedural nodes (they are "locked in"). */
if (n->tier && strcmp(n->tier, "Procedural") == 0) continue;
int64_t age_ms = now_ms - n->last_activated;
if (age_ms <= 0) continue;
double lambda = (n->temporal_decay_rate > 0.0) ? n->temporal_decay_rate
: ENGRAM_DECAY_LAMBDA;
double age_hours = (double)age_ms / 3600000.0;
double decay_factor = exp(-lambda * age_hours / ENGRAM_T_HALF_HOURS);
double new_salience = n->salience * decay_factor;
if (new_salience < 0.0) new_salience = 0.0;
if (new_salience != n->salience) {
n->salience = new_salience;
n->updated_at = now_ms;
updated++;
}
}
/* Prune low-salience Working nodes. Walk backwards to allow in-place
* removal without invalidating indices. */
for (int64_t i = g->node_count - 1; i >= 0; i--) {
EngramNode* n = &g->nodes[i];
if (n->salience >= 0.05) continue;
/* Only prune Working tier nodes — higher tiers are protected. */
if (!n->tier || strcmp(n->tier, "Working") != 0) continue;
/* Guard: skip nodes with no content. */
if (!n->content || !*n->content) continue;
/* Free node strings. */
free(n->id); free(n->content); free(n->node_type); free(n->label);
free(n->tier); free(n->tags); free(n->metadata);
/* Shift remaining nodes down. */
for (int64_t j = i + 1; j < g->node_count; j++) {
g->nodes[j - 1] = g->nodes[j];
}
g->node_count--;
memset(&g->nodes[g->node_count], 0, sizeof(EngramNode));
pruned++;
}
/* Remove dangling edges for pruned nodes (any edge whose endpoint no
* longer exists in the node list). */
if (pruned > 0) {
int64_t w = 0;
for (int64_t r = 0; r < g->edge_count; r++) {
EngramEdge* e = &g->edges[r];
int dangling = 0;
if (e->from_id && engram_find_node_index(e->from_id) < 0) dangling = 1;
if (e->to_id && engram_find_node_index(e->to_id) < 0) dangling = 1;
if (dangling) {
free(e->id); free(e->from_id); free(e->to_id);
free(e->relation); free(e->metadata);
} else {
if (w != r) g->edges[w] = g->edges[r];
w++;
/* Walk nodes array — skip any node whose id already exists */
const char* nodes_p = json_find_key(data, "nodes");
if (nodes_p) {
nodes_p = eg_skip_ws(nodes_p);
if (*nodes_p == '[') {
nodes_p++;
nodes_p = eg_skip_ws(nodes_p);
while (*nodes_p && *nodes_p != ']') {
if (*nodes_p != '{') { nodes_p++; continue; }
const char* end = json_skip_value(nodes_p);
size_t n = (size_t)(end - nodes_p);
char* obj = malloc(n + 1);
memcpy(obj, nodes_p, n); obj[n] = '\0';
char* nid = eg_get_str_field(obj, "id");
int already = (nid && *nid && engram_find_node(nid) != NULL);
free(nid);
if (!already) {
engram_grow_nodes();
EngramNode* nn = &g->nodes[g->node_count];
memset(nn, 0, sizeof(*nn));
nn->id = eg_get_str_field(obj, "id");
nn->content = eg_get_str_field(obj, "content");
nn->node_type = eg_get_str_field(obj, "node_type");
nn->label = eg_get_str_field(obj, "label");
nn->tier = eg_get_str_field(obj, "tier");
nn->tags = eg_get_str_field(obj, "tags");
nn->metadata = eg_get_str_field(obj, "metadata");
if (!nn->metadata || !*nn->metadata) { free(nn->metadata); nn->metadata = strdup("{}"); }
nn->salience = eg_get_num_field(obj, "salience");
nn->importance = eg_get_num_field(obj, "importance");
nn->confidence = eg_get_num_field(obj, "confidence");
nn->temporal_decay_rate = eg_get_num_field(obj, "temporal_decay_rate");
nn->activation_count = eg_get_int_field(obj, "activation_count");
nn->last_activated = eg_get_int_field(obj, "last_activated");
nn->created_at = eg_get_int_field(obj, "created_at");
nn->updated_at = eg_get_int_field(obj, "updated_at");
nn->background_activation = eg_get_num_field(obj, "background_activation");
nn->working_memory_weight = eg_get_num_field(obj, "working_memory_weight");
if (!isfinite(nn->working_memory_weight) || nn->working_memory_weight < 0.0 || nn->working_memory_weight > 1.0)
nn->working_memory_weight = 0.0; /* clamp corrupt snapshot values */
nn->suppression_count = (int32_t)eg_get_int_field(obj, "suppression_count");
if (json_find_key(obj, "layer_id")) {
nn->layer_id = (uint32_t)eg_get_int_field(obj, "layer_id");
} else {
nn->layer_id = ENGRAM_LAYER_DEFAULT;
}
g->node_count++;
added_nodes++;
}
free(obj);
nodes_p = end;
nodes_p = eg_skip_ws(nodes_p);
if (*nodes_p == ',') { nodes_p++; nodes_p = eg_skip_ws(nodes_p); }
}
}
g->edge_count = w;
}
char buf[128];
snprintf(buf, sizeof(buf),
"{\"ok\":true,\"updated\":%lld,\"pruned\":%lld}",
(long long)updated, (long long)pruned);
return el_wrap_str(el_strdup(buf));
/* Walk edges array — skip if (from_id, to_id, relation) already present */
const char* edges_p = json_find_key(data, "edges");
if (edges_p) {
edges_p = eg_skip_ws(edges_p);
if (*edges_p == '[') {
edges_p++;
edges_p = eg_skip_ws(edges_p);
while (*edges_p && *edges_p != ']') {
if (*edges_p != '{') { edges_p++; continue; }
const char* end = json_skip_value(edges_p);
size_t n = (size_t)(end - edges_p);
char* obj = malloc(n + 1);
memcpy(obj, edges_p, n); obj[n] = '\0';
char* efrom = eg_get_str_field(obj, "from_id");
char* eto = eg_get_str_field(obj, "to_id");
char* erel = eg_get_str_field(obj, "relation");
/* Check for duplicate by scanning existing edges */
int dup = 0;
if (efrom && eto && erel) {
for (int64_t ei = 0; ei < g->edge_count; ei++) {
EngramEdge* ex = &g->edges[ei];
if (ex->from_id && ex->to_id && ex->relation &&
strcmp(ex->from_id, efrom) == 0 &&
strcmp(ex->to_id, eto) == 0 &&
strcmp(ex->relation, erel) == 0) {
dup = 1; break;
}
}
}
if (!dup) {
engram_grow_edges();
EngramEdge* ee = &g->edges[g->edge_count];
memset(ee, 0, sizeof(*ee));
ee->id = eg_get_str_field(obj, "id");
ee->from_id = efrom ? efrom : strdup("");
ee->to_id = eto ? eto : strdup("");
ee->relation = erel ? erel : strdup("");
ee->metadata = eg_get_str_field(obj, "metadata");
if (!ee->metadata || !*ee->metadata) { free(ee->metadata); ee->metadata = strdup("{}"); }
ee->weight = eg_get_num_field(obj, "weight");
ee->confidence = eg_get_num_field(obj, "confidence");
ee->created_at = eg_get_int_field(obj, "created_at");
ee->updated_at = eg_get_int_field(obj, "updated_at");
ee->last_fired = eg_get_int_field(obj, "last_fired");
ee->inhibitory = (int)eg_get_int_field(obj, "inhibitory");
if (json_find_key(obj, "layer_id")) {
ee->layer_id = (uint32_t)eg_get_int_field(obj, "layer_id");
} else {
ee->layer_id = ENGRAM_LAYER_DEFAULT;
}
g->edge_count++;
/* NOTE: efrom/eto/erel ownership transferred to ee above */
efrom = NULL; eto = NULL; erel = NULL;
} else {
free(efrom); free(eto); free(erel);
}
free(obj);
edges_p = end;
edges_p = eg_skip_ws(edges_p);
if (*edges_p == ',') { edges_p++; edges_p = eg_skip_ws(edges_p); }
}
}
}
free(data);
return (el_val_t)added_nodes;
}
el_val_t engram_wm_count(void) {
EngramStore* g = engram_get();
int64_t count = 0;
for (int64_t i = 0; i < g->node_count; i++) {
if (g->nodes[i].working_memory_weight > 0.0) count++;
}
return (el_val_t)count;
}
/* Average working_memory_weight across all promoted nodes (wm > 0).
* Returns the float bit-pattern via el_from_float so EL can use it with
* float_to_str / float_gt. Returns 0.0 when no nodes are promoted.
* Useful in heartbeat ISEs to distinguish "many weak activations" (sparse
* graph, low avg) from "few strong activations" (dense subgraph, high avg).
* Added 2026-06-04 self-review for graph health observability. */
el_val_t engram_wm_avg_weight(void) {
EngramStore* g = engram_get();
double sum = 0.0;
int64_t count = 0;
for (int64_t i = 0; i < g->node_count; i++) {
double w = g->nodes[i].working_memory_weight;
/* Defensive guard: skip any corrupt/out-of-range values so a single
* bad snapshot node doesn't produce a garbage average (e.g. 1.77e+234). */
if (w > 0.0 && w <= 1.0 && isfinite(w)) { sum += w; count++; }
}
double avg = (count > 0) ? (sum / (double)count) : 0.0;
return el_from_float(avg);
}
/* engram_wm_top_json — return top N working-memory nodes (by wm weight) as a
* compact JSON array for ISE heartbeat reporting.
*
* Each element: {"label":"...","node_type":"...","tier":"...","wm":0.42}
*
* Purpose: the heartbeat ISE reports wm_active (count) and wm_avg_weight but
* gives zero visibility into WM *composition* which types/tiers are active.
* After long uptime every WM slot is in steady-state decay+re-promotion so
* wm_promotion ISEs never fire (they only fire on 0>0.1 transitions).
* This function fills the observability gap by snapshotting the current top-N
* WM nodes on every heartbeat. Inserted 2026-06-05 self-review. */
el_val_t engram_wm_top_json(el_val_t n_v) {
int64_t top_n = (int64_t)n_v;
if (top_n <= 0) top_n = 10;
if (top_n > 50) top_n = 50;
EngramStore* g = engram_get();
/* Collect indices of promoted nodes, excluding monitoring noise.
* InternalStateEvent nodes are system-observation artifacts they reflect
* what the daemon is doing, not what it knows. Including them in wm_top
* buries real knowledge (Memory, Knowledge, Belief nodes) under a wall of
* heartbeat/curiosity ISEs, making the heartbeat ISE useless for diagnosing
* WM composition. Filter them out here so wm_top always shows substantive
* content. (2026-06-07 self-review) */
int64_t* idx = malloc((size_t)(g->node_count + 1) * sizeof(int64_t));
if (!idx) return el_wrap_str(el_strdup("[]"));
int64_t mc = 0;
for (int64_t i = 0; i < g->node_count; i++) {
if (g->nodes[i].working_memory_weight > 0.0) {
const char* nt = g->nodes[i].node_type;
if (nt && strcmp(nt, "InternalStateEvent") == 0) continue;
idx[mc++] = i;
}
}
/* Insertion-sort descending by wm weight (mc is typically small). */
for (int64_t i = 1; i < mc; i++) {
int64_t key = idx[i];
double kw = g->nodes[key].working_memory_weight;
int64_t j = i;
while (j > 0 && g->nodes[idx[j-1]].working_memory_weight < kw) {
idx[j] = idx[j-1]; j--;
}
idx[j] = key;
}
int64_t emit = mc < top_n ? mc : top_n;
JsonBuf b; jb_init(&b);
jb_putc(&b, '[');
for (int64_t k = 0; k < emit; k++) {
EngramNode* n = &g->nodes[idx[k]];
if (k > 0) jb_putc(&b, ',');
jb_putc(&b, '{');
jb_puts(&b, "\"label\":");
jb_emit_escaped(&b, n->label ? n->label : "");
jb_puts(&b, ",\"node_type\":");
jb_emit_escaped(&b, n->node_type ? n->node_type : "");
jb_puts(&b, ",\"tier\":");
jb_emit_escaped(&b, n->tier ? n->tier : "");
char tmp[48];
snprintf(tmp, sizeof(tmp), ",\"wm\":%.3f", n->working_memory_weight);
jb_puts(&b, tmp);
jb_putc(&b, '}');
}
free(idx);
jb_putc(&b, ']');
return el_wrap_str(b.buf);
}
#ifdef HAVE_CURL
@@ -8757,7 +8883,7 @@ static el_val_t llm_provider_request(const char* url, const char* key,
}
}
static el_val_t llm_chain_call(const char* system_str, const char* user_str) {
static el_val_t llm_chain_call(const char* model_pref, const char* system_str, const char* user_str) {
char url_key[64], key_key[64], fmt_key[64], model_key[64];
for (int i = 0; i < LLM_MAX_PROVIDERS; i++) {
snprintf(url_key, sizeof(url_key), "NEURON_LLM_%d_URL", i);
@@ -8770,6 +8896,7 @@ static el_val_t llm_chain_call(const char* system_str, const char* user_str) {
const char* fmt_s = getenv(fmt_key);
int fmt = (fmt_s && strcmp(fmt_s, "anthropic") == 0) ? 1 : 0;
const char* model = getenv(model_key);
if (!model || !*model) model = model_pref; /* fall back to the caller-requested model */
fprintf(stderr, "[llm] trying provider %d (%s)\n", i, url);
el_val_t result = llm_provider_request(url, key, fmt, model, system_str, user_str);
const char* t = EL_CSTR(result);
@@ -8780,7 +8907,7 @@ static el_val_t llm_chain_call(const char* system_str, const char* user_str) {
const char* api_key = getenv("ANTHROPIC_API_KEY");
if (!api_key || !*api_key) return http_error_json("no LLM providers configured");
fprintf(stderr, "[llm] using legacy ANTHROPIC_API_KEY fallback\n");
return llm_provider_request(LLM_API_URL, api_key, 1, NULL, system_str, user_str);
return llm_provider_request(LLM_API_URL, api_key, 1, model_pref, system_str, user_str);
}
/* Legacy llm_request — kept for backward compat with agentic loop internals */
@@ -8844,14 +8971,16 @@ static el_val_t llm_extract_text(el_val_t resp_val) {
}
el_val_t llm_call(el_val_t model, el_val_t prompt) {
const char* m = EL_CSTR(model);
const char* u = EL_CSTR(prompt); if (!u) u = "";
return llm_chain_call(NULL, u);
return llm_chain_call(m, NULL, u);
}
el_val_t llm_call_system(el_val_t model, el_val_t system_prompt, el_val_t user_prompt) {
const char* m = EL_CSTR(model);
const char* s = EL_CSTR(system_prompt); if (!s) s = "";
const char* u = EL_CSTR(user_prompt); if (!u) u = "";
return llm_chain_call(s, u);
return llm_chain_call(m, s, u);
}
/* ── Tool registry for llm_call_agentic ─────────────────────────────────── */
+7
View File
@@ -176,6 +176,7 @@ el_val_t http_set_handler(el_val_t name);
* existing handlers (e.g. products/web/server.el): it dispatches with
* (method, path, body), hardcodes 200 OK, and auto-detects content type. */
el_val_t http_serve_v2(el_val_t port, el_val_t handler);
void http_serve_async(el_val_t port, el_val_t handler);
el_val_t http_set_handler_v2(el_val_t name);
/* Build an HTTP response envelope. `headers_json` should be a JSON object
@@ -638,6 +639,12 @@ el_val_t engram_list_layers_json(void);
* no nodes promoted to working memory. */
el_val_t engram_compile_layered_json(el_val_t intent, el_val_t depth);
/* ── Working memory ──────────────────────────────────────────────────────────*/
el_val_t engram_wm_count(void);
el_val_t engram_wm_avg_weight(void);
el_val_t engram_wm_top_json(el_val_t n);
el_val_t engram_load_merge(el_val_t path);
/* ── LLM (Anthropic API client) ─────────────────────────────────────────────
* All functions call https://api.anthropic.com/v1/messages with the API key
* from env ANTHROPIC_API_KEY. Default model when empty: claude-sonnet-4-5. */
-1
View File
@@ -2502,7 +2502,6 @@ fn builtin_arity(name: String) -> Int {
if str_eq(name, "http_post_with_headers") { return 3 }
if str_eq(name, "http_post_form_auth") { return 3 }
if str_eq(name, "http_serve") { return 2 }
if str_eq(name, "http_serve_async") { return 2 }
if str_eq(name, "http_set_handler") { return 1 }
// Seed primitives (__-prefix) runtime/el_seed.c
if str_eq(name, "__str_len") { return 1 }
+6 -21
View File
@@ -283,12 +283,6 @@ fn compile_module(src_path: String, out_dir: String, elc_bin: String, dry_run: B
}
exec_command("rm -f " + err_tmp)
// Strip capability-violation guard #error lines injected by elc when a
// module is compiled in isolation (utility context). These are safe to
// remove here: the entire binary is linked under the CGI entry-point
// declaration in soul.el, so the module-level guard is redundant.
exec_command("sed -i.bak '/^#error \"capability violation/d' " + c_out + " && rm -f " + c_out + ".bak")
// Move the generated .elh (written next to the source by elc) into
// out_dir so that #include "module.elh" lines in the generated .c
// files resolve correctly when cc is invoked with -I <out_dir>.
@@ -311,10 +305,6 @@ fn link_binary(c_files: [String], out_bin: String, runtime_path: String, out_dir
// prefix and add it if present (no-op on Linux where libssl is in /usr/lib).
let ossl_lib_flag: String = "$(brew --prefix openssl 2>/dev/null | xargs -I{} printf -- '-L{}/lib' 2>/dev/null || true)"
let ossl_inc_flag: String = "$(brew --prefix openssl 2>/dev/null | xargs -I{} printf -- '-I{}/include' 2>/dev/null || true)"
// liboqs (post-quantum crypto) present on macOS dev machines, not on CI
// Linux containers. Link -loqs only when the library is available.
let oqs_lib_flag: String = "$(brew --prefix liboqs 2>/dev/null | xargs -I{} printf -- '-L{}/lib -loqs' 2>/dev/null || true)"
let oqs_inc_flag: String = "$(brew --prefix liboqs 2>/dev/null | xargs -I{} printf -- '-I{}/include' 2>/dev/null || true)"
// Force-include the C-level master declarations header so every translation
// unit sees all cross-module function signatures. Handles packages (like ELP)
// where modules call each other without explicit El import statements.
@@ -322,7 +312,7 @@ fn link_binary(c_files: [String], out_bin: String, runtime_path: String, out_dir
let master_decls: String = out_dir + "/elp-c-decls.h"
let has_master: String = str_trim(exec_capture("test -f " + master_decls + " && echo yes || echo no"))
let include_flag: String = if str_eq(has_master, "yes") { "-include " + master_decls } else { "" }
let parts = native_list_append(parts, "cc -O2 -DHAVE_CURL " + bracket_flag + " " + ossl_inc_flag + " " + oqs_inc_flag + " " + include_flag + " -I " + dirname_of(runtime_path) + " -I " + out_dir)
let parts = native_list_append(parts, "cc -O2 " + bracket_flag + " " + ossl_inc_flag + " " + include_flag + " -I " + dirname_of(runtime_path) + " -I " + out_dir)
let i = 0
while i < n {
let f: String = native_list_get(c_files, i)
@@ -330,7 +320,7 @@ fn link_binary(c_files: [String], out_bin: String, runtime_path: String, out_dir
let i = i + 1
}
let parts = native_list_append(parts, runtime_path)
let parts = native_list_append(parts, ossl_lib_flag + " " + oqs_lib_flag + " -lcurl -lssl -lcrypto -lpthread -lm")
let parts = native_list_append(parts, ossl_lib_flag + " -lcurl -lssl -lcrypto -lpthread -lm")
let parts = native_list_append(parts, "-o " + out_bin)
let cmd: String = str_join(parts, " ")
println(" link " + out_bin)
@@ -442,23 +432,18 @@ fn main() -> Void {
exit(1)
}
// Link use only the entry-point .c file (which elc compiles as a
// monolithic unit, inlining all imports). Linking all module .c files
// together causes duplicate-symbol errors because each module's .c also
// inlines its full import tree.
let entry_c: String = out_dir + "/" + basename_noext(entry) + ".c"
let link_files: [String] = native_list_empty()
let link_files = native_list_append(link_files, entry_c)
// Append any extra C sources declared in the manifest (e.g. platform stubs)
let ei = 0
let en: Int = native_list_len(extra_c)
while ei < en {
let ec: String = native_list_get(extra_c, ei)
let link_files = native_list_append(link_files, ec)
let c_files = native_list_append(c_files, ec)
let ei = ei + 1
}
// Link
let out_bin: String = out_dir + "/" + pkg_name
let linked: Bool = link_binary(link_files, out_bin, runtime_path, out_dir, dry_run)
let linked: Bool = link_binary(c_files, out_bin, runtime_path, out_dir, dry_run)
if !linked {
println("elb: link failed")
exit(1)
-5
View File
@@ -1,5 +0,0 @@
CompileFlags:
Add:
- -I/opt/homebrew/Cellar/liboqs/0.15.0/include
- -I/opt/homebrew/opt/openssl@3/include
- -std=c11
File diff suppressed because it is too large Load Diff
@@ -143,7 +143,6 @@ el_val_t http_post_with_headers(el_val_t url, el_val_t body, el_val_t headers_m
el_val_t http_post_form_auth(el_val_t url, el_val_t form_body, el_val_t auth_header);
el_val_t http_delete(el_val_t url);
void http_serve(el_val_t port, el_val_t handler);
void http_serve_async(el_val_t port, el_val_t handler);
void http_set_handler(el_val_t name);
/* HTTP server v2 ─────────────────────────────────────────────────────────────
@@ -430,22 +429,10 @@ el_val_t str_to_float(el_val_t s);
el_val_t math_sqrt(el_val_t f);
el_val_t math_log(el_val_t f);
el_val_t math_ln(el_val_t f);
el_val_t math_exp(el_val_t f);
el_val_t math_sin(el_val_t f);
el_val_t math_cos(el_val_t f);
el_val_t math_pi(void);
/* ── Float arithmetic builtins (correct IEEE 754 via bit-cast round-trip) ─── */
el_val_t float_add(el_val_t a, el_val_t b);
el_val_t float_sub(el_val_t a, el_val_t b);
el_val_t float_mul(el_val_t a, el_val_t b);
el_val_t float_div(el_val_t a, el_val_t b);
el_val_t float_gt(el_val_t a, el_val_t b);
el_val_t float_lt(el_val_t a, el_val_t b);
el_val_t float_eq(el_val_t a, el_val_t b);
el_val_t float_gte(el_val_t a, el_val_t b);
el_val_t float_lte(el_val_t a, el_val_t b);
/* ── String additions ────────────────────────────────────────────────────── */
el_val_t str_index_of(el_val_t s, el_val_t sub);
@@ -505,7 +492,6 @@ el_val_t str_join(el_val_t list, el_val_t sep); /* alias of list_joi
el_val_t list_push(el_val_t list, el_val_t elem);
el_val_t list_push_front(el_val_t list, el_val_t elem);
el_val_t list_set(el_val_t list, el_val_t index, el_val_t value);
el_val_t list_join(el_val_t list, el_val_t sep);
el_val_t list_range(el_val_t start, el_val_t end);
@@ -603,10 +589,6 @@ el_val_t engram_edge_count(void);
el_val_t engram_activate(el_val_t query, el_val_t depth);
el_val_t engram_save(el_val_t path);
el_val_t engram_load(el_val_t path);
el_val_t engram_load_dir(el_val_t data_dir);
el_val_t engram_reindex_json(void);
el_val_t engram_write_binary_el(el_val_t path);
el_val_t engram_load_binary_el(el_val_t path);
/* JSON-string accessors — return pre-serialized JSON so HTTP handlers
* can pass results straight through without round-tripping ElList/ElMap
@@ -618,10 +600,6 @@ el_val_t engram_scan_nodes_by_type_json(el_val_t node_type, el_val_t limit, el_
el_val_t engram_neighbors_json(el_val_t node_id, el_val_t max_depth, el_val_t direction);
el_val_t engram_activate_json(el_val_t query, el_val_t depth);
el_val_t engram_stats_json(void);
el_val_t engram_wm_count(void);
el_val_t engram_wm_avg_weight(void); /* avg wm weight of promoted nodes; float bits */
el_val_t engram_wm_top_json(el_val_t n); /* top-N WM nodes by weight as compact JSON */
el_val_t engram_apply_decay_json(void);
el_val_t engram_list_layers_json(void);
/* engram_compile_layered_json — produce a prompt-ready text block split
* into "[LAYER 0 — STRUCTURAL]" (non-suppressible layers, sacred fire)
+43 -3
View File
@@ -6,15 +6,55 @@
//
// Dependencies: runtime/string.el, runtime/json.el
// --- Validation (defense in depth) ---
// el_val_t is an untyped machine word, so a wrong TYPE can't be caught here but a
// wrong VALUE can (a tier in the node_type slot, an empty/garbage string, an int, a
// path, a model name, a cgi id). Reject loudly instead of silently writing junk.
fn engram_valid_node_type(t: String) -> Bool {
return str_eq(t, "Memory") || str_eq(t, "Knowledge") || str_eq(t, "Belief")
|| str_eq(t, "Project") || str_eq(t, "Tag") || str_eq(t, "BacklogItem")
|| str_eq(t, "Artifact") || str_eq(t, "Conversation") || str_eq(t, "ExecutionContext")
|| str_eq(t, "InternalStateEvent") || str_eq(t, "Self") || str_eq(t, "Entity")
|| str_eq(t, "Process") || str_eq(t, "ConfigEntry") || str_eq(t, "Concept") || str_eq(t, "Imprint")
|| str_eq(t, "SessionSummary")
}
fn engram_valid_tier(t: String) -> Bool {
return str_eq(t, "Semantic") || str_eq(t, "Episodic") || str_eq(t, "Working")
|| str_eq(t, "Procedural") || str_eq(t, "Canonical") || str_eq(t, "Note") || str_eq(t, "Lesson")
}
// --- Node creation ---
fn engram_node(content: String, node_type: String, salience: Float) -> String {
if !engram_valid_node_type(node_type) {
__println("[engram] REJECTED node write — invalid node_type '" + node_type + "'")
return ""
}
return __engram_node(content, node_type, salience)
}
fn engram_node_full(content: String, nt: String, sal: Float, imp: Float,
source: String, lang: String, ts: Int, tags: String) -> String {
return __engram_node_full(content, nt, sal, imp, source, lang, ts, tags)
// Signature MUST match the C primitive __engram_node_full exactly (el_seed.h):
// (content, node_type, label, salience, importance, confidence, tier, tags)
// The previous wrapper declared a stale 8-arg schema with wrong names AND types
// (sal:Float at the label slot, ts:Int at the tier slot). Because el_val_t is an
// untyped machine word, the EL compiler coerced caller args to those wrong param
// types and then forwarded them BY POSITION into the C function so tier received
// an int, importance/confidence received strings, label received a float, etc.
// That is the field-corruption bug. Match the contract 1:1 no coercion, no reorder.
fn engram_node_full(content: String, node_type: String, label: String,
salience: Float, importance: Float, confidence: Float,
tier: String, tags: String) -> String {
if !engram_valid_node_type(node_type) {
__println("[engram] REJECTED node write — invalid node_type '" + node_type + "' (label=" + label + ")")
return ""
}
if !engram_valid_tier(tier) {
__println("[engram] REJECTED node write — invalid tier '" + tier + "' (node_type=" + node_type + ", label=" + label + ")")
return ""
}
return __engram_node_full(content, node_type, label, salience, importance, confidence, tier, tags)
}
// --- Node retrieval ---