triton6564685/gbrain
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
e3f704229b9314ee4c799347622c5b2d95a9859e
gbrain/CLAUDE.md
Garry Tan 8b3c24c891 v0.20.0 feat: extract BrainBench to sibling gbrain-evals repo (#195) 
* fix(link-extraction): v0.10.5 drive works_at + advises accuracy on rich prose

Extends inferLinkType patterns to cover rich-prose phrasings that miss with
v0.10.4 regexes. Targets the residuals called out in TODOS.md: works_at at
58% type accuracy, advises at 41%.

WORKS_AT_RE additions:
- Rank-prefixed: "senior engineer at", "staff engineer at", "principal/lead"
- Discipline-prefixed: "backend/frontend/full-stack/ML/data/security engineer at"
- Possessive time: "his/her/their/my time at"
- Leadership beyond "leads engineering": "heads up X at", "manages engineering at",
  "runs product at", "leads the [team] at"
- Role nouns: "role at", "position at", "tenure as", "stint as"
- Promotion patterns: "promoted to staff/senior/principal at"

ADVISES_RE additions:
- Advisory capacity: "in an advisory capacity", "advisory engagement/partnership/contract"
- "as an advisor": "joined as an advisor", "serves as technical advisor"
- Prefixed advisor nouns: "strategic/technical/security/product/industry advisor to|at"
- Consulting: "consults for", "consulting role at|with"

New EMPLOYEE_ROLE_RE page-level prior: fires when the page describes the subject
as an employee (senior/staff/principal engineer, director, VP, CTO/CEO/CFO) at
some company. Biases outbound company refs toward works_at when per-edge context
is possessive or narrative without an explicit work verb. Scoped to person -> company
links only. Precedence: investor > advisor > employee (investors often hold board
seats which would otherwise mis-classify as advise/works_at).

ADVISOR_ROLE_RE broadened from "full-time/professional/advises multiple" to catch
any page that self-identifies the subject as an advisor ("is an advisor",
"serves as advisor", possessive "her advisory work/role/engagement").

Tests: 65 pass (16 new v0.10.5 coverage tests + 4 regression guards against
v0.10.4 tightenings). Templated benchmark still 88.9% type_accuracy (10/10 on
works_at and advises). Rich-prose measurement requires the multi-axis report
upgrade (next commit) to validate retroactively.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): type-accuracy runner on rich-prose corpus + wire into all.ts

New Category 2 in BrainBench: per-link-type accuracy measured directly on the
240-page rich-prose world-v1 corpus. Distinct from Cat 1's retrieval metrics,
this measures whether inferLinkType() correctly classifies extracted edges
when the prose varies (the 58% works_at and 41% advises residuals that v0.10.5
regexes targeted).

How it works:
  1. Loads all pages from eval/data/world-v1/
  2. Derives GOLD expected edges from each page's _facts metadata
     (founders → founded, investors → invested_in, advisors → advises,
      employees → works_at, attendees → attended, primary_affiliation +
      role drives person-page outbound type)
  3. Runs extractPageLinks() on each page → INFERRED edges
  4. Per (from, to) pair, compares inferred type vs gold type
  5. Emits per-link-type table: correct / mistyped / missed / spurious +
     type accuracy + recall + precision + strict F1 (triple match)
  6. Full confusion matrix rows=gold, cols=inferred

v0.10.5 validation on 240-page corpus (up from pre-v0.10.5 baselines):
  - works_at:    58%  → 100.0%   (+42 pts) — 10/10 correct, 0 mistyped
  - advises:     41%  → 88.2%    (+47 pts) — 15/17 correct
  - attended:    —    → 100.0%   131/134 recall
  - founded:    100%  → 100.0%   40/40
  - invested_in: 89%  → 92.0%    69/75
  - Overall:    88.5% → 95.7%    type accuracy (conditional on edge found)

Strict F1 overall: 53.7%. Lower because the _facts-based gold set only
captures core relationships; rich prose extracts many peripheral mentions
(190 spurious "mentions" edges) that aren't bugs but are correctly-typed
prose references without a _facts counterpart. Spurious counts are signal
for future type-precision tuning, not failure.

Wired into eval/runner/all.ts as Cat 2 so every full benchmark run includes
the rich-prose type accuracy table alongside retrieval metrics.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): Phase 2 adapter interface + EXT-1 ripgrep+BM25 baseline

Phase 2 credibility unlock: BrainBench now compares gbrain to external
baselines on the same corpus and queries. Transforms the benchmark from
internal ablation ("gbrain-graph beats gbrain-grep") to category comparison
("gbrain-graph beats classic BM25 by 32 pts P@5"). This is the #1 fix
from the 4-review arc — addresses Codex's core critique that v1's
before/after was self-referential.

Added:
  eval/runner/types.ts                      — Adapter interface (v1.1 spec)
  eval/runner/adapters/ripgrep-bm25.ts      — EXT-1 classic IR baseline
  eval/runner/adapters/ripgrep-bm25.test.ts — 11 unit tests, all pass
  eval/runner/multi-adapter.ts              — side-by-side scorer

Adapter interface (eng pass 2 spec):
  - Thin 3-method Strategy: init(rawPages, config), query(q, state), snapshot(state)
  - BrainState is opaque to runner (never inspected)
  - Raw pages passed in-memory; gold/ never crosses adapter boundary
    (structural ingestion-boundary enforcement)
  - PoisonDisposition enum reserved for future poison-resistance scoring

EXT-1 ripgrep+BM25:
  - Classic Lucene-variant IDF + k1/b tuned at standard 1.5/0.75
  - Title tokens double-weighted for entity-page slug-match bias
  - Stopword filter, alphanumeric tokenization, stable lexicographic tie-break
  - Pure in-memory inverted index — no external deps, ~100 LOC core

First side-by-side results on 240-page rich-prose corpus, 145 relational queries:

| Adapter       | P@5    | R@5    | Correct top-5 |
|---------------|--------|--------|---------------|
| gbrain-after  | 49.1%  | 97.9%  | 248/261       |
| ripgrep-bm25  | 17.1%  | 62.4%  | 124/261       |
| Delta         | +32.0  | +35.5  | +124          |

gbrain-after is the hybrid graph+grep config from PR #188. Ripgrep+BM25 is
a genuinely strong classic-IR baseline (BM25 is what Lucene/Elasticsearch
ship). gbrain's ~+32-point lead on relational queries reflects real work
by the knowledge graph layer: typed links + traversePaths surface the
correct answers in top-K that BM25 only pulls in via partial-text overlap.

Next in Phase 2: EXT-2 vector-only RAG + EXT-3 hybrid-without-graph
adapters. Both plug into the same Adapter interface.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): Phase 2 EXT-2 vector-only RAG adapter

Second external baseline for BrainBench. Pure cosine-similarity ranking
using the SAME text-embedding-3-large model gbrain uses internally —
apples-to-apples on the embedding layer so any gbrain lead reflects the
graph + hybrid fusion, not a better embedder.

Files:
  eval/runner/adapters/vector-only.ts      ~130 LOC
  eval/runner/adapters/vector-only.test.ts 6 unit tests (cosine math)

Design:
  - One vector per page (title + compiled_truth + timeline, capped 8K chars).
  - No chunking (intentional; chunked vector RAG would be EXT-2b later).
  - No keyword fallback (that's EXT-3 hybrid-without-graph).
  - Embeddings in batches of 50 via existing src/core/embedding.ts (retry+backoff).
  - Cost on 240 pages: ~$0.02/run.

Three-adapter side-by-side on 240-page rich-prose corpus, 145 relational queries:

| Adapter       | P@5    | R@5    | Correct top-5 |
|---------------|--------|--------|---------------|
| gbrain-after  | 49.1%  | 97.9%  | 248/261       |
| ripgrep-bm25  | 17.1%  | 62.4%  | 124/261       |
| vector-only   | 10.8%  | 40.7%  |  78/261       |

Interesting finding: vector-only scores WORSE than BM25 on relational queries
like "Who invested in X?" — exact entity match matters more than semantic
similarity for these templates. BM25 nails the entity-name term; vector-only
returns topically-similar-but-not-mentioning pages. This is the known failure
mode of pure-vector RAG on precise relational/identity queries. Real-world
vector RAG systems always add keyword fallback; EXT-3 (hybrid-without-graph)
will be that fairer comparator.

gbrain's lead widens in vector-only comparison: +38.4 pts P@5, +57.2 pts R@5.
The graph layer is doing the heavy lifting for relational traversal; pure
vector RAG can't express "traverse 'attended' edges from this meeting page."

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): Phase 2 EXT-3 hybrid-without-graph adapter — graph isolated

Third and closest-to-gbrain external baseline. Runs gbrain's full hybrid
search (vector + keyword + RRF fusion + dedup) WITHOUT the knowledge-graph
layer. Same engine, same embedder, same chunking, same hybrid fusion —
only traversePaths + typed-link extraction turned off.

This is the decisive comparator for "does the knowledge graph do useful
work?" Same everything-else, only graph differs. Any lead gbrain-after has
over EXT-3 is 100% attributable to the graph layer.

Files:
  eval/runner/adapters/hybrid-nograph.ts   — ~110 LOC

Implementation:
  - New PGLiteEngine per run; auto_link set to 'false' (belt).
  - importFromContent() used instead of bare putPage() so chunks +
    embeddings get populated (hybridSearch needs them).
  - NO runExtract() call — typed links/timeline stay empty (suspenders).
  - hybridSearch(engine, q.text) answers every query. Aggregate chunks
    to page-level by best chunk score.

FOUR-adapter side-by-side on 240-page rich-prose corpus, 145 relational queries:

| Adapter         | P@5    | R@5    | Correct/Gold |
|-----------------|--------|--------|--------------|
| gbrain-after    | 49.1%  | 97.9%  | 248/261      |
| hybrid-nograph  | 17.8%  | 65.1%  | 129/261      |
| ripgrep-bm25    | 17.1%  | 62.4%  | 124/261      |
| vector-only     | 10.8%  | 40.7%  |  78/261      |

The headline delta nobody can hand-wave away:
  gbrain-after → hybrid-nograph  = +31.4 P@5, +32.9 R@5
  hybrid-nograph → ripgrep-bm25  = +0.7 P@5,  +2.7 R@5

Hybrid search (vector+keyword+RRF) over pure BM25 gains ~1 point. The
knowledge graph layer over hybrid gains ~31 points. The graph is doing
the work; adding it to a retrieval stack is what actually moves the needle
on relational queries. The vector/keyword/BM25 debate is a footnote.

Timing: hybrid-nograph init is ~2 min (embeds 240 pages once); query loop
is fast. gbrain-after is ~1.5s total because traversePaths doesn't need
embeddings. Runs at ~$0.02 Opus-equivalent in embedding cost.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): Phase 2 query validator + Tier 5 Fuzzy + Tier 5.5 synthetic + N=5 tolerance bands

Closes multiple Phase 2 items in one commit since they form a cohesive
package: query schema enforcement + new query tiers + per-query-set
statistical rigor.

Added:
  eval/runner/queries/validator.ts               — hand-rolled Query schema validator
  eval/runner/queries/validator.test.ts          — 24 unit tests, all pass
  eval/runner/queries/tier5-fuzzy.ts             — 30 hand-authored Tier 5 Fuzzy/Vibe queries
  eval/runner/queries/tier5_5-synthetic.ts       — 50 SYNTHETIC-labeled outsider-style queries (author: "synthetic-outsider-v1")
  eval/runner/queries/index.ts                   — aggregator + validateAll()

Modified:
  eval/runner/multi-adapter.ts                   — N=5 runs per adapter (BRAINBENCH_N override), page-order shuffle, mean±stddev reporting

Query validator (hand-rolled, no zod dep to match gbrain codebase style):
  - Temporal verb regex enforces as_of_date (per eng pass 2 spec):
    /\\b(is|was|were|current|now|at the time|during|as of|when did)\\b/i
  - Validates tier enum, expected_output_type enum, gold shape per type
  - gold.relevant must be non-empty slug[] for cited-source-pages queries
  - abstention requires gold.expected_abstention === true
  - externally-authored tier requires author field
  - batch validation catches duplicate IDs

Tier 5 Fuzzy/Vibe (30 queries, hand-authored):
  - Vague recall: "Someone who was a senior engineer at a biotech company..."
  - Trait-based: "The engineer who pushed back on microservices"
  - Cultural/epithet: "Who is known as a 'systems builder' in security?"
  - Abstention bait: "Which Layer 1 project did the crypto guy leave?" (prose
    mentions but never names; good systems abstain)
  - Addresses Codex's circularity critique — vague queries where graph-heavy
    systems shouldn't inherently win.

Tier 5.5 Synthetic Outsider (50 queries, AI-authored placeholder):
  - Clearly labeled author: "synthetic-outsider-v1"
  - Phrasing variety not in the 4 template families:
    * fragment style ("crypto founder Goldman Sachs background")
    * polite/natural ("Can you pull up what we have on...")
    * comparison ("What is the difference between X and Y?")
    * follow-up ("And who else advises Orbit Labs?")
    * typos/misspellings ("adam lopez bioinformatcis")
    * similarity ("Find me someone like Alice Davis...")
    * imperative ("Pull up Alice Davis")
  - Real Tier 5.5 from outside researchers supersedes synthetic via
    PRs to eval/external-authors/ (docs ship in follow-up commit).

N=5 tolerance bands:
  - Default N=5, override via BRAINBENCH_N env var (e.g. BRAINBENCH_N=1 for dev loops)
  - Per-run seeded Fisher-Yates shuffle of page ingest order (LCG seed = run_idx+1)
  - Surfaces order-dependent adapter bugs (tie-break-by-first-seen etc.)
  - Reports mean ± sample-stddev per metric
  - "stddev = 0" is honest signal that the adapter is deterministic, not a bug.
    LLM-judge metrics (future) will naturally produce non-zero stddev.

Validation: all 80 Tier 5 + 5.5 queries pass validateAll(). 24 validator
unit tests pass.

Next commit: world.html contributor explorer (Phase 3).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* feat(eval): Phase 3 world.html explorer + eval:* CLI surface

Contributor DX magical moment. Static HTML explorer renders the full
canonical world (240 entities) as an explorable tree, opens in any browser,
zero install. Every string HTML-entity-encoded (XSS-safe — direct vuln
class per eng pass 2, confidence 9/10).

Added:
  eval/generators/world-html.ts         — renderer (~240 LOC; single-file
                                          HTML with inline CSS + minimal JS)
  eval/generators/world-html.test.ts    — 16 tests (XSS + rendering correctness)
  eval/cli/world-view.ts                — render + open in default browser
  eval/cli/query-validate.ts            — CLI wrapper for queries/validator
  eval/cli/query-new.ts                 — scaffold a query template

Modified:
  package.json                          — 7 new eval:* scripts
  .gitignore                            — ignore generated world.html

package.json scripts shipped:
  bun run test:eval                 all eval unit tests (57 pass)
  bun run eval:run                  full 4-adapter N=5 side-by-side
  bun run eval:run:dev              N=1 fast dev iteration
  bun run eval:world:view           render world.html + open in browser
  bun run eval:world:render         render only (CI-friendly, --no-open)
  bun run eval:query:validate       validate built-in T5+T5.5 (or a file path)
  bun run eval:query:new            scaffold a new Query JSON template
  bun run eval:type-accuracy        per-link-type accuracy report

XSS safety:
  escapeHtml() encodes the 5 critical chars (& < > " '). Tested directly
  with representative Opus-generated attacks:
    <img src=x onerror=alert('xss')>  → &lt;img src=x onerror=alert(&#39;xss&#39;)&gt;
    <script>fetch('/steal')</script>  → &lt;script&gt;fetch(&#39;/steal&#39;)&lt;/script&gt;
  Ledger metadata (generated_at, model) also escaped — covers the less
  obvious attack surface where Opus could emit tag-like content into the
  metadata file.

world.html structure:
  - Left rail: entities grouped by type with counts (companies, people,
    meetings, concepts), alphabetical within type
  - Right pane: per-entity cards with title + slug + compiled_truth +
    timeline + canonical _facts as collapsed JSON
  - URL fragment deep-links (#people/alice-chen)
  - Sticky rail on desktop; responsive stack on mobile
  - Vanilla JS for active-link highlighting on scroll (no framework)

Generated file: ~1MB for 240 entities (full prose). Gitignored; rebuild
with `bun run eval:world:view`. Regeneration is ~50ms.

Contributor TTHW (Tier 5.5 query authoring):
  1. bun run eval:world:view                         # see entities
  2. bun run eval:query:new --tier externally-authored --author "@me"
  3. edit template with real slug + query text
  4. bun run eval:query:validate path/to/file.json
  5. submit PR

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(eval): Phase 3 contributor docs + CI workflow for eval/ tests

Ships the contributor-onboarding surface promised in the plan. With this
commit, external researchers have a self-serve path from clone to PR in
under 5 minutes.

Added:
  eval/README.md                                — 5-minute quickstart,
                                                  directory map, methodology
                                                  one-pager, adapter scorecard
  eval/CONTRIBUTING.md                          — three contributor paths:
                                                    1. Write Tier 5.5 queries
                                                    2. Submit an external adapter
                                                    3. Reproduce a scorecard
  eval/RUNBOOK.md                               — operational troubleshooting:
                                                  generation failures, runner
                                                  failures, query validation,
                                                  world.html rendering, CI
  eval/CREDITS.md                               — contributor attribution
                                                  (synthetic-outsider-v1 labeled
                                                  as placeholder; real submissions
                                                  land here)
  .github/PULL_REQUEST_TEMPLATE/tier5-queries.md — structured PR template
                                                  for Tier 5.5 submissions
  .github/workflows/eval-tests.yml              — CI: validates queries,
                                                  runs all eval unit tests,
                                                  renders world.html on every PR
                                                  touching eval/** or
                                                  src/core/link-extraction.ts

CI scope (intentionally narrow):
  - Triggers on paths: eval/**, src/core/link-extraction.ts, src/core/search/**
  - Runs: bun run eval:query:validate (80 queries), test:eval (57 tests),
          eval:world:render (smoke-test the HTML renderer)
  - Pinned actions by commit SHA (matches existing .github/workflows/test.yml)
  - Zero API calls — all Opus/OpenAI paths stubbed or skipped in unit tests
  - Fast: ~30s total wall clock

Contributor TTHW (clone → first merged PR):
  - Path 1 (Tier 5.5 queries): ~5 min
  - Path 2 (external adapter): ~30 min for a simple adapter
  - Path 3 (reproduce scorecard): ~15 min wall clock (N=5 run)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(eval): teardown PGLite engines so bun run eval:run exits 0

The multi-adapter runner left PGLite engines alive after each run.
GbrainAfterAdapter and HybridNoGraphAdapter both instantiate a
PGLiteEngine in init() but never disconnect it; Bun's shutdown path
exits with code 99 when embedded-Postgres workers outlive main().

Added optional `teardown?(state)` to the Adapter interface, implemented
it on both engine-backed adapters, and call it from scoreOneRun after
the N=5 loop. ripgrep-bm25 and vector-only hold no DB resources and
don't need a teardown.

Verified: gbrain-after, hybrid-nograph, ripgrep-bm25, vector-only all
exit 0 at N=1. Full test:eval passes (57 tests). No metric change.

* docs(bench): 2026-04-19 multi-adapter scorecard

Reproducibility run of the 4-adapter side-by-side at commit b81373d
(branch garrytan/gbrain-evals). N=5, 240-page corpus, 145 relational
queries from world-v1.

Headline: gbrain-after 49.1% P@5 / 97.9% R@5. hybrid-nograph 17.8% /
65.1%. ripgrep-bm25 17.1% / 62.4%. vector-only 10.8% / 40.7%. All
adapters deterministic (stddev = 0 across the 5 runs per adapter).

Matches the scorecard in eval/README.md byte-for-byte for the three
deterministic adapters; hybrid-nograph matches within tolerance bands.

* docs(bench): 2026-04-19 gbrain v0.11.1 vs v0.12.1 regression comparison

Runs the same eval harness against two gbrain src/ trees on the same
240-page corpus and 145 queries. Patches the v0.11 copy's gbrain-after
adapter to use getLinks/getBacklinks (v0.11 has no traversePaths)
with identical direction+linkType semantics.

gbrain-after P@5 22.1% -> 49.1% (+27 pts); R@5 54.6% -> 97.9% (+43
pts); correct-in-top-5 99 -> 248 (+149). hybrid-nograph flat at 17.8%
/ 65.1% on both (v0.12 didn't touch hybridSearch / chunking).

Driver is extraction quality, not graph presence: v0.12 emits 499
typed links (v0.11: 136, x3.7) and 2,208 timeline entries (v0.11: 27,
x82) on the same 240 pages. Sharpens the April-18 "graph layer does
the work" claim -- on v0.11 that architecture only beat hybrid-nograph
by 4.3 points; the 31-point lead in the multi-adapter scorecard comes
from graph + high-quality extract in combination.

* feat(eval): BrainBench v1 portable JSON schemas + gold templates

Adds the v1→v2 contract boundary for BrainBench. 6 JSON schemas at
eval/schemas/ pin the shape of every artifact a stack must emit to be
scorable: corpus-manifest, public-probe (PublicQuery with gold stripped),
tool-schema (12 read + 3 dry_run tools, 32K tool-output cap), transcript,
scorecard (N ∈ {1, 5, 10}), evidence-contract (structured judge input).

8 gold file templates at eval/data/gold/ scaffold the sealed qrels,
contradictions, poison items, and citation labels. Empty-but-valid
skeletons; Day 3b fills them with real content once the amara-life-v1
corpus generates.

48 tests validate schema syntax, $schema/$id/title/type headers,
round-trip stability, and cross-schema coherence (new Page types in
manifest enum, tool counts, token cap, N enum).

When v2 ports to Python + Inspect AI + Docker, these schemas are the
boundary. Same fixtures, same tool contracts, zero rework.

* feat(eval): amara-life-v1 skeleton + Page.type enum for email/slack/cal/note

Deterministic procedural generator for the twin-amara-lite fictional-life
corpus (BrainBench v1 Cat 5/8/9/11 target). 15 contacts picked from
world-v1, 50 emails + 300 Slack messages across 4 channels + 20 calendar
events + 8 meeting transcripts + 40 first-person notes. Mulberry32 PRNG
gives byte-identical output under reseed.

Plants 10 contradictions + 5 stale facts + 5 poison items + 3 implicit
preferences at deterministic positions. Fixture_ids are unique across the
corpus so gold/contradictions.json + gold/poison.json + gold/implicit-
preferences.json can cross-reference by stable ID.

PageType extended in both src/core/types.ts and eval/runner/types.ts to
include email | slack | calendar-event | note (+ meeting on the production
side). src/core/markdown.ts inferType() heuristics updated for the new
one-slash slug prefixes (emails/em-NNNN, slack/sl-NNNN, cal/evt-NNNN,
notes/YYYY-MM-DD-topic, meeting/mtg-NNNN).

17 tests cover counts (50/300/20/8/40), perturbation counts (exact
10/5/5/3), seed determinism + divergence, slug regex conformance (matches
eval/runner/queries/validator.ts:131 one-slash rule), unique fixture_ids,
amara-in-every-email invariant, calendar dtstart < dtend, and Amara-is-
attendee on every meeting.

* feat(eval): amara-life-gen.ts with structured cache key + $20 cost gate

Opus prose expansion of the amara-life-v1 skeleton. Per-item structured
cache key = sha256({schema_version, template_id, template_hash, model_id,
model_params, seed, item_spec_hash}). Prompt-template tweak changes
template_hash; only those items regenerate. Schema bump changes
schema_version; everything invalidates cleanly. Interrupted runs resume
from the last cached item; zero re-spend.

Cost-gated at $20 hard-stop with Anthropic input/output pricing tracking.
Dry-run mode (--dry-run) executes the full pipeline with stub bodies for
smoke-testing the I/O layout without LLM spend. --max N caps items per
type for debugging. --force ignores cache.

Writes per-format outputs under eval/data/amara-life-v1/:
  inbox/emails.jsonl (one email per line with body_text appended)
  slack/messages.jsonl (one message per line with text appended)
  calendar.ics (RFC-5545 VEVENT format, templated — no LLM)
  meetings/<id>.md (transcript with YAML frontmatter)
  notes/<YYYY-MM-DD-topic>.md (first-person journal)
  docs/*.md (6 reference docs, templated — no LLM)
  corpus-manifest.json (per eval/schemas/corpus-manifest.schema.json,
    including per-item content_sha256 and generator_cache_key)

Perturbation hints (contradiction, stale-fact, poison, implicit-
preference) flow through the prompt so Opus weaves the specific claim
into each item's body. Poison items are hand-crafted to include
paraphrased prompt-injection attempts (not literal 'IGNORE ALL
PREVIOUS' — defense is the structured-evidence judge contract at
Day 5, not regex redaction).

New package.json scripts:
  eval:generate-amara-life       # real run (~$12 Opus estimated)
  eval:generate-amara-life:dry   # smoke test, zero spend

test:eval extended to include test/eval/. 10 cache-key tests cover
determinism, invalidation across every field of the key, canonical JSON
stability under object-key reorder, and per-skeleton-item spec-hash
uniqueness (50 distinct hashes for 50 distinct emails).

* chore: bump version and changelog (v0.15.0)

Resets package.json from stale 0.13.1 to 0.15.0 (matches VERSION).
v0.14.0 shipped with the stale package.json version; this sync catches
that up and moves to v0.15.0 in one step.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* docs: update CLAUDE.md + README + eval/README for v0.15.0 BrainBench

CLAUDE.md: adds a full BrainBench section to the Key Files list — 14 new
entries covering eval/README.md, multi-adapter.ts, types.ts (with new
PublicPage/PublicQuery), adapters/, queries/, type-accuracy.ts,
adversarial.ts, all.ts, world.ts/gen.ts, world-html.ts, amara-life.ts,
amara-life-gen.ts, schemas/, data/world-v1/, data/gold/,
data/amara-life-v1/, docs/benchmarks/, and test/eval/. Adds 3 new
test/eval/ lines to the unit-tests catalog.

eval/README.md: file tree updated to reflect v0.15 additions —
data/amara-life-v1/, data/gold/, schemas/, generators/amara-life.ts +
amara-life-gen.ts, runner/all.ts + adversarial.ts.

README.md: updates hero benchmark numbers (L7 intro + L353 mid-page)
from v0.10.5 PR #188 numbers (R@5 83→95, P@5 39→45) to current v0.12.1
4-adapter numbers (P@5 49.1% · R@5 97.9% · +31.4 pts vs hybrid-nograph).
Adds the v0.11→v0.12 regression comparison as the secondary reference.
Deeper-section tables (L422+) labeled "BrainBench v1 (PR #188)" are
preserved as historical data.

CHANGELOG is untouched — /ship already wrote the v0.15.0 entry.
TODOS.md is untouched — Cat 5/6/8/9/11 remain open (only foundations
shipped in v0.15.0; Cat runners ship in v1 Complete follow-ups).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 4 — pdf-parse + flight-recorder + tool-bridge (dry_run + expand:false)

Three infrastructure modules for BrainBench v1 Complete Cats 5/8/9/11.

**eval/runner/loaders/pdf.ts** — Thin pdf-parse wrapper. Lazy import keeps
pdf-parse out of the module-load path (avoids library debug-mode side
effects). Size cap (50MB default), encryption detection, structured error
classes (PdfEncryptedError, PdfTooLargeError, PdfParseError). Only Cat 11
multimodal will import this; production bundle never sees pdf-parse.

**eval/runner/tool-bridge.ts** — Maps 12 read-only operations from
src/core/operations.ts to Anthropic tool definitions + adds 3 dry_run write
tools. Three structural invariants enforced:

  1. No hidden LLM calls. `operations.query` defaults expand=true which
     routes through expansion.ts → Haiku. Bridge strips `expand` from the
     query tool's input schema AND executor hard-sets expand:false. Zero
     nested Haiku calls in any agent trace.

  2. Mutating ops throw ForbiddenOpError. put_page, add_link, delete_page,
     etc. are rejected by name. Agents record intent via dry_run_put_page /
     dry_run_add_link / dry_run_add_timeline_entry which persist to the
     flight-recorder without mutating the engine. This is how Cat 8's
     back_link_compliance + citation_format metrics measure anything with
     a read-only tool surface.

  3. Poison tagged by the bridge, not the judge. Every tool result is
     scanned for slugs matching gold/poison.json fixtures. Matched
     fixture_ids flow into tool_call_summary.saw_poison_items for the
     structured-evidence judge contract. Judge never reads raw tool
     output — Section-3 defense against paraphrased prompt injections
     (poison payloads never reach the judge model at all).

32K-token cap (~128K chars) with "…[truncated]" suffix.

**eval/runner/recorder.ts** — Per-run flight-recorder bundle emitter. Full
6-artifact bundle (transcript.md, brain-export.json, entity-graph.json,
citations.json, scorecard.json, judge-notes.md) when the adapter provides
an AdapterExport; 3-artifact fallback (transcript + scorecard +
judge-notes) otherwise. Atomic writes via tmp+rename. Collision-safe:
duplicate directory names get incremental -2, -3 suffix. `safeStringify`
handles circular references without throwing and JSON-serializes
Float32Array embeddings.

**package.json:** adds pdf-parse@2.4.5 as a devDependency. Scoped to eval/
use only; production gbrain binary unaffected.

**Tests:** 63 new — 30 tool-bridge, 21 recorder, 12 pdf-loader. All pass.
Fake engine uses a Proxy with `__default__` fallback so poison-matching
tests don't have to mock the exact engine method name that each operation
calls (some route via searchKeyword, others via getPage — proxy handles
both uniformly).

Total eval suite now: 132 pass, 0 fail, 923 expect() calls.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 5 — agent adapter + judge with structured evidence contract

Two modules that together wire Cat 8 / Cat 9 / Cat 5 end-to-end scoring.

**eval/runner/judge.ts** — Haiku 4.5 via tool-use `score_answer`. Input is
the structured JudgeEvidence contract (fix #16 from the plan's codex
review): probe + final_answer_text + evidence_refs + tool_call_summary +
ground_truth_pages + rubric. Raw tool output NEVER reaches the judge —
that's the Section-3 defense against paraphrased prompt-injection payloads
in gold/poison.json.

Retry policy: one retry on malformed tool_use response. If the second
attempt is still malformed, score the probe as `judge_failed` (all scores
0, verdict=fail) so the run still completes.

Aggregation: weighted mean across rubric criteria. Canonical thresholds
(pass ≥3.5, partial 2.5-3.5, fail <2.5) — judge can propose a verdict but
the computed verdict from the weighted mean is what the scorecard records.
This prevents the model from inflating or deflating its own verdict.

Score values are clamped to 0-5 on parse even if the model returns out of
range. `assertNoRawToolOutput(evidence)` is a regression guard that
returns the list of forbidden fields (tool_result, raw_transcript, etc.)
if any leak into the evidence contract.

**eval/runner/adapters/claude-sonnet-with-tools.ts** — The agent adapter.
Implements `Adapter` interface minimally: `init()` spins up PGLite and
seeds it, `query()` throws because the adapter is Cat 8/9-only and emits
a final-answer text, not a RankedDoc[]. Retrieval scorecard stays at 4
adapters.

`runAgentLoop(probeId, text, state, config)` drives the multi-turn loop:
Sonnet → tool_use → tool-bridge.executeTool → tool_result → back to
Sonnet. Turn cap 10. max_tokens 1024. System prompt (brain-first iron
law, citation format, amara context) is cached via cache_control.
Exponential backoff on rate-limit errors (1s, 2s, 4s).

Emits a `Transcript` per eval/schemas/transcript.schema.json — consumed
directly by recorder.ts for the flight-recorder bundle.

`brain_first_ordering` classifies Cat 8's flagship metric: did the agent
call search/get_page BEFORE producing the final answer? The `no_brain_calls`
case (agent answers from general knowledge without ever hitting the brain)
is the compliance failure to surface.

ForbiddenOpError + UnknownToolError from the bridge are caught in the
agent loop and surfaced as tool_result with is_error=true — keeps the
loop going and preserves full audit trail for the judge.

**Tests (35 new):** judge (23) — happy path, retry, fallback, evidence
contract sanitization, rendered prompt does not contain raw tool_result
text, verdict thresholds, score clamping, weighted mean with mixed
weights, parseToolUse rejects malformed input. agent-adapter (12) —
Adapter.query() throws, init() seeds PGLite, end-to-end tool loop with
stubbed Sonnet, turn cap exhaustion, mutating-op rejection surfaces as
tool_result error, extractSlugs regex.

All 12 agent tests take ~23s because PGLite runs 13 schema migrations per
test; the alternative of shared-engine-across-tests was rejected so each
test is isolated.

Total eval suite now: 167 pass, 0 fail.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 6 — adversarial-injections + Cat 6 prose-scale + Cat 11 multi-modal

Three modules that together cover BrainBench v1 Cat 6 (prose-scale
extraction fidelity) and Cat 11 (multi-modal ingest fidelity).

**eval/runner/adversarial-injections.ts** — 6 deterministic content
transforms shared by Cat 10 (adversarial.ts, 22 hand-crafted cases) and
Cat 6 (prose-scale variants). Each injection produces a modified content
string + a structured GoldDelta describing what the extractor MUST and
MUST NOT produce. Kinds:
  - code_fence_leak — fake [X](people/fake) inside ``` fence, must NOT extract
  - inline_code_slug — `people/fake` in backticks, must NOT extract
  - substring_collision — "SamAI" near real `people/sam`, exactly one link
  - ambiguous_role — "works with" vs "works at", downgrade type to mentions
  - prose_only_mention — strip markdown link syntax, bare name → mentions only
  - multi_entity_sentence — pack 4+ entities into one clause, extract all

Mulberry32 PRNG keeps variant generation deterministic under fixed seed.
Codex flagged the original plan's wording ("extract injection engine from
adversarial.ts") as overstated — adversarial.ts is a static case list,
not a reusable engine. This module is NEW code.

**eval/runner/cat6-prose-scale.ts** — Runner. Loads world-v1, applies all
6 injection kinds to sampled base pages (default 50 variants per kind ×
6 kinds = 300 variants), runs extractPageLinks on each, compares to gold
delta. Emits per-kind + overall metrics (precision, recall, F1,
code_fence_leak_rate, substring_fp_rate, pages_with_links_coverage,
mean_links_per_page). **v1 verdict is always "baseline_only"** — no
gating threshold per codex fix #9 (current extractor residuals make
>0.80 unreachable; v1 records a baseline, regression guard triggers on
drop below it).

**eval/runner/cat11-multimodal.ts** — PDF + HTML + audio runners.
Fixtures load from eval/data/multimodal/<modality>/fixtures.json
manifests; each modality skips gracefully when manifest missing or
(audio) when neither GROQ_API_KEY nor OPENAI_API_KEY is set. Metrics:
  - PDF: char-level similarity via Levenshtein + optional entity_recall
  - HTML: word-recall over normalized tokens (multiset semantics)
  - Audio: WER (word error rate) via Levenshtein on word sequences
Fixtures are NOT committed; a future eval:fetch-multimodal script will
download them hash-verified from public sources (arXiv CC-licensed
papers, Wikipedia CC-BY-SA, Common Voice CC0).

Injectable audio transcriber (`opts.transcribe`) means tests don't need
GROQ/OpenAI keys — stubbed transcriptions exercise the WER math path
directly.

**Tests (60 new):** adversarial-injections (19) — per-kind assertions +
dispatcher coverage + slug regex conformance; cat6 (12) — variant
determinism, scoreVariant shape, aggregate per-kind + overall metrics,
corpus resolver slug rules; cat11 (29) — charSimilarity / wordRecall /
wer math, htmlToText strips scripts + decodes entities, HTML modality
with real fixtures, audio modality gracefully skips without key + uses
stub transcriber correctly.

All 60 tests pass in 48ms + 41ms.
Total eval suite now: 227 pass, 0 fail.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 7 — Cat 5 provenance runner + structured classify_claim judge

**eval/runner/cat5-provenance.ts** — BrainBench Cat 5 scoring. Samples
claims from gbrain brain-export and classifies each against its source
material via a dedicated Haiku judge (classify_claim tool with a
three-label enum: supported | unsupported | over-generalized).

Separate from judge.ts by design: Cat 5 is a single three-way
classification per claim, not a weighted rubric. Rather than overload
judge.ts with a mode switch, Cat 5 has its own tool definition
(CLASSIFY_CLAIM_TOOL) and prompt. The retry-once pattern, $20 cost gate
semantics, and structured parsing are mirrored from judge.ts so failures
look the same across Cats.

Metric: `citation_accuracy` = fraction where predicted label equals
gold expected_label. Threshold (informational): >0.90 per design-doc
METRICS.md. v1 ships with `enableThreshold: false` so the verdict is
always baseline_only — we don't have hand-authored gold claims yet, and
codex flagged that threshold gating should wait until the amara-life-v1
corpus + gold file authoring lands in Day 3b.

runCat5 uses a bounded-concurrency worker pool (default 4) to respect
Haiku rate limits across 100+ claim batches. Evidence pages are looked
up by slug from a caller-provided pagesBySlug map — missing pages don't
crash, they just pass an empty source list to the judge (correct
behavior for genuinely unsupported claims).

**Tests (23):** classifyClaim happy/retry/fallback paths with stubbed
Haiku, aggregate accuracy math, threshold gating (pass/fail vs
baseline_only), runCat5 concurrency + missing-page handling,
renderClaimPrompt embeds claim + sources correctly, parseClassification
rejects invalid enum values + plain-text responses.

Total eval suite now: 250 pass, 0 fail.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 8 — Cat 8 skill compliance + Cat 9 end-to-end workflows

**eval/runner/cat8-skill-compliance.ts** — Deterministic, judge-free Cat 8
scoring. Replays inbound signals through the agent adapter (Day 5) and
extracts four iron-law metrics directly from the tool-bridge state:

  - brain_first_compliance: agent called search/get_page BEFORE producing
    its final answer. Non-compliance = hallucinating from general knowledge.
  - back_link_compliance: every dry_run_put_page intent has at least one
    markdown [Name](slug) back-link in its compiled_truth.
  - citation_format: timeline entries use canonical `- **YYYY-MM-DD** |
    Source — Summary`; long final answers cite at least one slug.
  - tier_escalation: simple probes use light tooling (≥1 brain call);
    complex probes require ≥2 brain calls or a dry_run write when
    expects_dry_run_write is set.

No judge call required — everything is computable from
`tool_bridge_state.made_dry_run_writes` + `count_by_tool` + final_answer
regex. Fast, deterministic, reproducible.

Bounded concurrency (p-limit style) worker pool at default 4 to keep
Sonnet rate limits comfortable across 100-probe batches.

**eval/runner/cat9-workflows.ts** — Rubric-graded Cat 9. 5 canonical
workflows (meeting_ingestion, email_to_brain, daily_task_prep, briefing,
sync) × ~10 scenarios each. Each scenario runs through the agent adapter,
then judge.ts scores the answer against a per-scenario rubric.

`buildEvidence(scenario, agentResult, pagesBySlug)` composes the
JudgeEvidence contract: resolves ground_truth_slugs to full
GroundTruthPage[] from a slug-map, pulls tool_call_summary directly from
tool_bridge_state (no raw tool_result content — Section-3 defense),
attaches rubric from the scenario.

Per-workflow rollup: each workflow gets its own pass_rate so the verdict
can fail one workflow without failing the whole Cat. Overall verdict
requires every populated workflow's pass_rate ≥ threshold (default 0.80)
when enableThreshold=true.

Both Cats default to verdict=baseline_only in v1 per codex fix #9: real
thresholds return after 10-probe Haiku-vs-hand-score calibration (κ > 0.7)
runs against the Day 3b amara-life-v1 corpus.

**Tests (23):** Cat 8 per-metric scorer unit tests covering every branch
(brain_first ordering, back-link compliance on mixed writes, long vs
short answer citation requirement, tier escalation for simple/complex/
writey probes, finalAnswerCiteCount dedups across syntaxes). Cat 9
buildEvidence contract shape — evidence_refs flow from agent, missing
slugs skip gracefully, no raw_transcript/tool_result leakage to judge.
Cat 9 runCat9 integration with stubbed agent + mixed-verdict judge
produces fractional pass rates correctly.

Total eval suite now: 273 pass, 0 fail.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 9 — sealed qrels via PublicPage + PublicQuery at adapter boundary

Codex fixes #1, #2, #3 from the plan's outside-voice review. Enforcement
shifts from SOFT-VIA-TYPE-COMMENT to SOFT-VIA-SANITIZED-OBJECT. Hard
enforcement via process isolation waits for BrainBench v2 Docker sandbox.

**eval/runner/types.ts** additions:
  - `PublicPage = Pick<Page, 'slug' | 'type' | 'title' | 'compiled_truth' |
    'timeline'>` — the exact 5 fields adapters should see. No _facts.
    No frontmatter (a known hiding spot for accidental gold leaks).
  - `sanitizePage(p: Page): PublicPage` — returns a NEW object with the 5
    fields only. Cannot be bypassed by `(page as any)._facts` because the
    field does not exist on the sanitized object.
  - `PublicQuery = Omit<Query, 'gold'>` — strips the gold field.
  - `sanitizeQuery(q: Query): PublicQuery` — enumerates public fields
    explicitly (not spread+delete) so no prototype weirdness leaves gold
    reachable.

**eval/runner/multi-adapter.ts** — scoreOneRun now calls sanitizePage /
sanitizeQuery before passing to adapter.init / adapter.query. The scorer
retains the full Query shape (including gold.relevant) for precision /
recall computation. Adapter signatures unchanged — the sealing is at the
OBJECT level, not the type level. This keeps existing adapters
(ripgrep-bm25, vector-only, hybrid-nograph, gbrain-after) binary-compatible.
Verified: no existing adapter reads q.gold or page._facts, so the change
is safe without further adapter updates.

**test/eval/sealed-qrels.test.ts** (17 tests):
  - sanitizePage strips _facts + frontmatter + arbitrary hidden keys
  - Output has exactly the 5 public keys (deep introspection)
  - Proxy tripwire simulates a malicious adapter: any access to _facts or
    gold throws `sealed-qrels violation`
  - sanitizeQuery retains optional fields (as_of_date, tags, author,
    acceptable_variants, known_failure_modes) but omits undefined ones
  - Honest documentation of the seal's limits: filesystem bypass and
    Proxy attacks would still work in v1; Docker isolation (v2) is the
    real enforcement

Every existing eval test still passes (273 before + 17 sealed-qrels = 290).

Total eval suite now: 290 pass, 0 fail.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* feat(eval): Day 10 — all.ts rewrite + llm-budget + BrainBench N tiers

Final wiring of BrainBench v1 Complete. all.ts now orchestrates the full
Cat catalog (1-12) via a mix of subprocess dispatch (Cats 1, 2, 3, 4, 6,
7, 10, 11, 12 — standalone runners with CLI entry points) and
programmatic invocation (Cats 5, 8, 9 — require runtime inputs that
can't come via CLI flags). Subprocess Cats run concurrently under a
p-limit(2) bound to cap peak memory around ~800MB (two PGLite instances
at ~400MB each).

Cats 5/8/9 show as "programmatic" in the report with a one-line
reference to their `runCatN({...})` harness API. They're deliberately
skipped from the master runner because their inputs (claim catalog,
probe catalog, scenario catalog, pre-seeded agent state, evidence
pagesBySlug) are task-specific and assembled at the caller.

**eval/runner/all.ts** — rewritten:
  - CATEGORIES is a tagged union of SubprocessCategory | ProgrammaticCategory
  - runCatSubprocess spawns Bun with pipe'd stdout/stderr, 10-min timeout
    per Cat (124 exit + SIGTERM on timeout; no hung subprocesses)
  - runConcurrently is a bounded worker pool preserving input order
  - buildReport emits the full markdown with per-Cat elapsed times,
    migration-noise filter, and a separate programmatic-only section
  - Honors BRAINBENCH_N (1/5/10 for smoke/iteration/published),
    BRAINBENCH_CONCURRENCY (default 2),
    BRAINBENCH_LLM_CONCURRENCY (default 4, consumed by llm-budget)

**eval/runner/llm-budget.ts** — shared LLM rate-limit semaphore. A full
N=10 published scorecard makes ~900 Anthropic calls (150 Cat 8/9 probes
× N=10 + 100 Cat 5 claims × N=10). Without coordination, concurrent
adapters trigger 429s on per-minute limits.

  - LlmBudget class: acquireSlot/releaseSlot + withLlmSlot(fn) wrapper
    that releases on success AND throw (try/finally)
  - getDefaultLlmBudget() singleton reads BRAINBENCH_LLM_CONCURRENCY,
    falls back to 4 on missing/garbage values
  - capacity enforced ≥1 (rejects 0/negative)
  - Double-release is a no-op (guards against upstream double-call bugs)
  - Active + waiting counts exposed for observability / tests

**package.json** scripts:
  - eval:brainbench           — default N=5 iteration
  - eval:brainbench:smoke     — N=1 for fast iteration
  - eval:brainbench:published — N=10 for committed baselines
  - eval:cat6 / eval:cat11    — individual new subprocess Cats

**Tests (24):** CATEGORIES catalog enforces the exact Cat-number partition
(subprocess: 1,2,3,4,6,7,10,11,12; programmatic: 5,8,9). runConcurrently
respects the cap (observable via peak in-flight counter), preserves input
order under non-uniform delays, handles empty input. LlmBudget enforces
capacity, releases on throw, honors env var, rejects 0/negative.
buildReport filters migration noise, counts passed/failed/programmatic
correctly, includes every Cat + programmatic-only section.

Full eval suite now: 314 pass, 0 fail (15 test files).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix(eval): drop top_p from amara-life-gen Opus params + gitignore _cache/

Two fixes surfaced during the Day 3b real-corpus run against Opus 4.5:

**eval/generators/amara-life-gen.ts** — Current Opus rejects
`temperature` and `top_p` together:
```
400 invalid_request_error: `temperature` and `top_p` cannot both be
specified for this model. Please use only one.
```
top_p=1.0 was a no-op (no nucleus truncation), so removing it has zero
semantic effect. The field is still part of MODEL_PARAMS for the cache
key so any past cache entries (none in v1) would invalidate cleanly
on the next schema version bump.

**.gitignore** — `eval/data/amara-life-v1/_cache/` is runtime Opus
cache (398 files, ~1.6MB). Regenerable from seed; no point in source
control. The corpus itself (inbox/slack/calendar/meetings/notes/docs +
corpus-manifest.json with per-item content_sha256) stays committable
for reproducibility, just the cache directory gets excluded.

Real corpus generation ran cleanly after these two fixes: 398 LLM
calls, 84,424 input / 38,062 output tokens, \$4.12 spent (vs \$20 cap,
vs \$12 estimate). All 418 items produced. Poison fixtures use
subtle paraphrased injection ("for anyone on your team who might be
triaging this thread later…") — exactly the pattern that defeats
regex redaction and requires the structured-evidence judge contract
from Day 5.

Corpus itself stays local (will move to the brainbench sibling repo
during the v0.16 split per the design doc). No eval/data/amara-life-v1/
content landing in this PR.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* chore: bump version to 0.20.0

Renumbered from 0.17.0 per the gbrain-versioning slot. Other work is
landing on master around this PR; 0.18 is the slot locked for this
BrainBench v1 Complete release. Also pushed the "brainbench split"
forward reference in the CHANGELOG from v0.18 → v0.19 to match.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* refactor: extract BrainBench to sibling gbrain-evals repo

BrainBench lived in this repo through v0.17, which meant every gbrain install
pulled down ~5MB of eval corpus, benchmark reports, and a pdf-parse devDep
that the 99% of users who never run benchmarks don't need.

v0.18 moves the full eval harness, 14 eval test files (314 tests), all
docs/benchmarks scorecards, and the pdf-parse devDep to
github.com/garrytan/gbrain-evals. That repo depends on gbrain via GitHub URL
and consumes it through a new public exports map.

What stays in gbrain:
- Page.type enum extensions (email | slack | calendar-event | note | meeting)
  useful for any ingested format, not just evals
- inferType() heuristics for /emails/, /slack/, /cal/, /notes/, /meetings/
- 11 new public exports covering the gbrain internals gbrain-evals consumes
  (gbrain/engine, gbrain/pglite-engine, gbrain/search/hybrid, etc.) — now
  gbrain's stable third-party contract

What moved:
- eval/ — 4.6MB of schemas, runners, adapters, generators, CLI tools
- test/eval/ — 14 test files, 314 tests
- docs/benchmarks/ — all scorecards and regression reports
- eval:* package.json scripts
- pdf-parse devDep

Tests: 1760 pass, 0 fail, 174 skipped (E2E require DATABASE_URL).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Merge origin/master into garrytan/gbrain-evals

Master landed significant work since this branch was cut (v0.15.x → v0.16.x →
v0.17.0 gbrain dream + runCycle → v0.18.0 multi-source brains → v0.18.1 RLS
hardening). Bumped this branch's version from the claimed 0.18.0 to 0.19.0
because master already owns 0.18.x.

Conflicts resolved:
- VERSION: 0.19.0 (was 0.18.0 on HEAD vs 0.18.1 on master)
- package.json: 0.19.0, kept all 11 eval-facing exports, merged master's
  typescript devDep + postinstall script + test script (typecheck added)
- src/core/types.ts: union of both PageType additions. Master had added
  `meeting | note`; this branch added `email | slack | calendar-event`
  for inbox/chat/calendar ingest. Final enum carries all five.
- CHANGELOG.md: renumbered the BrainBench-extraction entry to 0.19.0 and
  placed it above master's 0.18.1 RLS entry. Tweaked copy ("In v0.17 it
  lived inside this repo" → "Previously it lived inside this repo") to
  stop implying a specific version that never shipped.
- CLAUDE.md: adjusted "BrainBench in a sibling repo" heading from
  (v0.18+) → (v0.19+).
- docs/benchmarks/2026-04-18-minions-vs-openclaw-production.md:
  resolved modify-vs-delete conflict in favor of delete (the extraction).
- scripts/llms-config.ts: dropped the docs/benchmarks/ entry (directory
  no longer exists here; lives in gbrain-evals).
- llms.txt / llms-full.txt: regenerated after the config change.
- bun.lock: accepted master's (master already dropped pdf-parse as a
  drive-by; aligned with our removal).

Tests: 2094 pass, 236 skip, 18 fail. Spot-checked failures — build-llms,
dream, orphans tests all pass in isolation. Failures reproduce only under
full-suite parallel load and are pre-existing master flakiness (matches the
graph-quality flake noted in the earlier summary). Not merge-introduced.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* chore: bump to v0.20.0

Master is now at v0.18.2 (migration hardening + RLS + multi-source brains).
BrainBench extraction ships as v0.20.0 to leave v0.19 free for any in-flight
work on other branches.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* ci: remove eval-tests workflow (moved to gbrain-evals)

The Eval tests workflow ran `bun run eval:query:validate`, `test:eval`, and
`eval:world:render` — all three scripts moved to the gbrain-evals repo when
BrainBench was extracted in v0.20.0. The workflow has been failing on master
since the split because the scripts no longer exist here.

Eval CI now runs from gbrain-evals's own workflows.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* fix(tests): bump PGLite hook timeouts to 60s for parallel-load stability

Six test files spin up PGLite + 20 migrations + git repos in beforeEach/
beforeAll hooks. Under 136-way parallel test file execution, bun's default
5s hook timeout wasn't enough, producing 18 flaky failures that only
reproduced under full-suite parallel load (all 6 files passed in isolation).

Root cause: PGLite.create() + initSchema() takes ~3-5s under idle load, but
under 136 concurrent WASM instantiations the OS thrashes and hooks stall
well past 5s. The bunfig.toml `timeout = 60_000` applies to TESTS, not HOOKS
— bun requires per-hook timeouts as the third beforeEach/beforeAll argument.

Files touched (hook timeouts added, no test logic changed):
- test/dream.test.ts           — 5 describe blocks × before/afterEach
- test/orphans.test.ts         — 1 beforeEach + afterEach
- test/core/cycle.test.ts      — shared beforeAll + afterAll
- test/brain-allowlist.test.ts — beforeAll + afterAll
- test/extract-db.test.ts      — beforeAll + afterAll
- test/multi-source-integration.test.ts — beforeAll + afterAll

Results: 2317 pass / 0 fail (was 2253 pass / 18 fail).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* test: coverage for inferType() BrainBench corpus dirs

Closes the 1 gap surfaced by Step 7 coverage audit. 9 table-driven
assertions covering the new Page.type branches:
  emails/*.md, email/*.md       -> 'email'
  slack/*.md                    -> 'slack'
  cal/*.md, calendar/*.md       -> 'calendar-event'
  notes/*.md, note/*.md         -> 'note'
  meetings/*.md, meeting/*.md   -> 'meeting'

The fixtures use realistic paths from the amara-life-v1 corpus in the
sibling gbrain-evals repo (em-0001, sl-0037, evt-0042, mtg-0003) so the
test doubles as a contract check between the two repos.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs(TODOS): mark BrainBench Cats 5/6/8/9/11 + v0.10.5 inferLinkType as completed

All five BrainBench categories shipped in v0.20.0 (to the gbrain-evals
sibling repo). v0.10.5 inferLinkType regex expansion shipped in-tree.

Remaining P1 BrainBench work: Cat 1+2 at full scale (2-3K pages) —
currently 240 pages in world-v1 corpus.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs: sync CLAUDE.md + polish CHANGELOG voice for v0.20.0

CLAUDE.md: add v0.19 commands to key-files list (skillify, skillpack,
routing-eval, filing-audit, skill-manifest, resolver-filenames);
add 8 new test files + openclaw-reference-compat E2E to test index;
repoint the release-summary template's benchmark source from
`docs/benchmarks/[latest].md` to `gbrain-evals/docs/benchmarks/` since
those files now live in the sibling repo.

CHANGELOG voice polish for v0.20.0: replace em dashes with periods,
parens, or ellipses per project style guide. No content changes.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* docs: regenerate llms-full.txt after CLAUDE.md + CHANGELOG edits (fixes CI)

The v0.20.0 doc-sync commit (9e567bb) added 7 new v0.19 modules to the
CLAUDE.md Key Files index and polished CHANGELOG voice. Both are
includeInFull: true inputs to llms-full.txt but the generator wasn't
re-run, so the drift-detection guard (test/build-llms.test.ts) failed CI.

One-line fix: regenerate. No content changes beyond what the two source
docs already carry.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-24 00:08:54 -07:00

821 lines
73 KiB
Markdown
Raw Blame History

# CLAUDE.md
GBrain is a personal knowledge brain and GStack mod for agent platforms. Pluggable
engines: PGLite (embedded Postgres via WASM, zero-config default) or Postgres + pgvector
+ hybrid search in a managed Supabase instance. `gbrain init` defaults to PGLite;
suggests Supabase for 1000+ files. GStack teaches agents how to code. GBrain teaches
agents everything else: brain ops, signal detection, content ingestion, enrichment,
cron scheduling, reports, identity, and access control.
## Architecture
Contract-first: `src/core/operations.ts` defines ~41 shared operations (adds `find_orphans` in v0.12.3). CLI and MCP
server are both generated from this single source. Engine factory (`src/core/engine-factory.ts`)
dynamically imports the configured engine (`'pglite'` or `'postgres'`). Skills are fat
markdown files (tool-agnostic, work with both CLI and plugin contexts).
**Trust boundary:** `OperationContext.remote` distinguishes trusted local CLI callers
(`remote: false` set by `src/cli.ts`) from untrusted agent-facing callers
(`remote: true` set by `src/mcp/server.ts`). Security-sensitive operations like
`file_upload` tighten filesystem confinement when `remote=true` and default to
strict behavior when unset.
## Key files
- `src/core/operations.ts` — Contract-first operation definitions (the foundation). Also exports upload validators: `validateUploadPath`, `validatePageSlug`, `validateFilename`. `OperationContext.remote` flags untrusted callers.
- `src/core/engine.ts` — Pluggable engine interface (BrainEngine). `clampSearchLimit(limit, default, cap)` takes an explicit cap so per-operation caps can be tighter than `MAX_SEARCH_LIMIT`. Exports `LinkBatchInput` / `TimelineBatchInput` for the v0.12.1 bulk-insert API (`addLinksBatch` / `addTimelineEntriesBatch`). As of v0.13.1, `BrainEngine` has a `readonly kind: 'postgres' | 'pglite'` discriminator so migrations (`src/core/migrate.ts`) and other consumers can branch on engine without `instanceof` + dynamic imports.
- `src/core/engine-factory.ts` — Engine factory with dynamic imports (`'pglite'` | `'postgres'`)
- `src/core/pglite-engine.ts` — PGLite (embedded Postgres 17.5 via WASM) implementation, all 40 BrainEngine methods. `addLinksBatch` / `addTimelineEntriesBatch` use multi-row `unnest()` with manual `$N` placeholders. As of v0.13.1, `connect()` wraps `PGlite.create()` in a try/catch that emits an actionable error naming the macOS 26.3 WASM bug (#223) and pointing at `gbrain doctor`; the lock is released on failure so the next process can retry cleanly.
- `src/core/pglite-schema.ts` — PGLite-specific DDL (pgvector, pg_trgm, triggers)
- `src/core/postgres-engine.ts` — Postgres + pgvector implementation (Supabase / self-hosted). `addLinksBatch` / `addTimelineEntriesBatch` use `INSERT ... SELECT FROM unnest($1::text[], ...) JOIN pages ON CONFLICT DO NOTHING RETURNING 1` — 4-5 array params regardless of batch size, sidesteps the 65535-parameter cap. As of v0.12.3, `searchKeyword` / `searchVector` scope `statement_timeout` via `sql.begin` + `SET LOCAL` so the GUC dies with the transaction instead of leaking across the pooled postgres.js connection (contributed by @garagon). `getEmbeddingsByChunkIds` uses `tryParseEmbedding` so one corrupt row skips+warns instead of killing the query.
- `src/core/utils.ts` — Shared SQL utilities extracted from postgres-engine.ts. Exports `parseEmbedding(value)` (throws on unknown input, used by migration + ingest paths where data integrity matters) and as of v0.12.3 `tryParseEmbedding(value)` (returns `null` + warns once per process, used by search/rescore paths where availability matters more than strictness).
- `src/core/db.ts` — Connection management, schema initialization
- `src/commands/migrate-engine.ts` — Bidirectional engine migration (`gbrain migrate --to supabase/pglite`)
- `src/core/import-file.ts` — importFromFile + importFromContent (chunk + embed + tags)
- `src/core/sync.ts` — Pure sync functions (manifest parsing, filtering, slug conversion)
- `src/core/storage.ts` — Pluggable storage interface (S3, Supabase Storage, local)
- `src/core/supabase-admin.ts` — Supabase admin API (project discovery, pgvector check)
- `src/core/file-resolver.ts` — File resolution with fallback chain (local -> .redirect.yaml -> .redirect -> .supabase)
- `src/core/chunkers/` — 3-tier chunking (recursive, semantic, LLM-guided)
- `src/core/search/` — Hybrid search: vector + keyword + RRF + multi-query expansion + dedup
- `src/core/search/intent.ts` — Query intent classifier (entity/temporal/event/general → auto-selects detail level)
- `src/core/search/eval.ts` — Retrieval eval harness: P@k, R@k, MRR, nDCG@k metrics + runEval() orchestrator
- `src/commands/eval.ts` — `gbrain eval` command: single-run table + A/B config comparison
- `src/core/embedding.ts` — OpenAI text-embedding-3-large, batch, retry, backoff
- `src/core/check-resolvable.ts` — Resolver validation: reachability, MECE overlap, DRY checks, structured fix objects. v0.14.1: `CROSS_CUTTING_PATTERNS.conventions` is an array (notability gate accepts both `conventions/quality.md` and `_brain-filing-rules.md`). New `extractDelegationTargets()` parses `> **Convention:**`, `> **Filing rule:**`, and inline backtick references. DRY suppression is proximity-based via `DRY_PROXIMITY_LINES = 40`.
- `src/core/repo-root.ts` — Shared `findRepoRoot(startDir?)` (v0.16.4): walks up from `startDir` (default `process.cwd()`) looking for `skills/RESOLVER.md`. Zero-dependency module imported by both `doctor.ts` and `check-resolvable.ts`. Parameterized `startDir` makes tests hermetic.
- `src/commands/check-resolvable.ts` — Standalone CLI wrapper (v0.16.4) over `checkResolvable()`. Exports `parseFlags`, `resolveSkillsDir`, `DEFERRED`, `runCheckResolvable`. Exit rule: **1 on any issue (warnings OR errors)**, stricter than doctor's `ok` flag — honors README:259. Stable JSON envelope `{ok, skillsDir, report, autoFix, deferred, error, message}` — same shape on success and error paths. `--fix` path runs `autoFixDryViolations` BEFORE `checkResolvable` (same ordering as doctor). `scripts/skillify-check.ts` subprocess-calls `gbrain check-resolvable --json` (cached per process) and fails loud on binary-missing — no silent false-pass. **v0.19:** AGENTS.md workspaces now resolve natively (see `src/core/resolver-filenames.ts`) — gbrain inspects the 107-skill OpenClaw deployment whether the routing file is `RESOLVER.md` or `AGENTS.md`. `DEFERRED[]` is empty — Checks 5 + 6 shipped as real code, not issue URLs.
- `src/core/resolver-filenames.ts` (v0.19) — central list of accepted routing filenames (`RESOLVER.md`, `AGENTS.md`). Shared by `findRepoRoot`, `check-resolvable`, and skillpack install so every code path walks the same fallback chain.
- `src/commands/skillify.ts` + `src/core/skillify/{generator,templates}.ts` (v0.19) — `gbrain skillify scaffold <name>` creates all stubs for a new skill in one command: SKILL.md, script, tests, routing-eval.jsonl, resolver entry, filing-rules pointer. `gbrain skillify check <script>` runs the 10-step checklist (LLM evals, routing evals, check-resolvable gate, filing audit) against a candidate skill before it lands.
- `src/commands/skillify-check.ts` (v0.19) — `gbrain skillpack-check` agent-readable health report. Exit 0/1/2 for CI pipeline gating; JSON for debugging. Wraps `check-resolvable --json`, `doctor --json`, and migration ledger into one payload so agents can decide whether a human action is required.
- `src/commands/skillpack.ts` + `src/core/skillpack/{bundle,installer}.ts` (v0.19) — `gbrain skillpack install` drops gbrain's curated 25-skill bundle into a host workspace, managed-block style. Never clobbers local edits; tracks a skill manifest so subsequent `install --update` diffs cleanly. Bundle builder (`skillpack/bundle.ts`) packages the set from `skills/` into a versioned payload.
- `src/core/skill-manifest.ts` (v0.19) — parser for `skill-manifest.json` records. Used by skillpack installer to detect drift between the shipped bundle and the user's local edits, so updates merge instead of overwriting.
- `src/commands/routing-eval.ts` + `src/core/routing-eval.ts` (v0.19) — `gbrain routing-eval` catches user phrasings that route to the wrong skill. Reads `skills/<name>/routing-eval.jsonl` fixtures (`{intent, expected_skill, ambiguous_with?}`). Structural layer runs in `check-resolvable` by default (zero API cost); `--llm` opts into a Haiku tie-break layer for CI. False positives surface before users hit them.
- `src/core/filing-audit.ts` + `skills/_brain-filing-rules.json` (v0.19) — Check 6 of `check-resolvable`. Parses new `writes_pages:` / `writes_to:` frontmatter on skills and audits their filing claims against the filing-rules JSON. Warning-only in v0.19, upgrades to error in v0.20.
- `src/core/dry-fix.ts` — `gbrain doctor --fix` engine. `autoFixDryViolations(fixes, {dryRun})` rewrites inlined rules to `> **Convention:** see [path](path).` callouts via three shape-aware expanders (bullet / blockquote / paragraph). Five guards: working-tree-dirty (`getWorkingTreeStatus()` returns 3-state `'clean' | 'dirty' | 'not_a_repo'`), no-git-backup, inside-code-fence, already-delegated (40-line proximity, consistent with detector), ambiguous-multi-match, block-is-callout. `execFileSync` array args (no shell — no injection surface). EOF newline preserved.
- `src/core/backoff.ts` — Adaptive load-aware throttling: CPU/memory checks, exponential backoff, active hours multiplier
- `src/core/fail-improve.ts` — Deterministic-first, LLM-fallback loop with JSONL failure logging and auto-test generation
- `src/core/transcription.ts` — Audio transcription: Groq Whisper (default), OpenAI fallback, ffmpeg segmentation for >25MB
- `src/core/enrichment-service.ts` — Global enrichment service: entity slug generation, tier auto-escalation, batch throttling
- `src/core/data-research.ts` — Recipe validation, field extraction (MRR/ARR regex), dedup, tracker parsing, HTML stripping
- `src/commands/extract.ts` — `gbrain extract links|timeline|all [--source fs|db]`: batch link/timeline extraction. fs walks markdown files, db walks pages from the engine (mutation-immune snapshot iteration; use this for live brains with no local checkout). As of v0.12.1 there is no in-memory dedup pre-load — candidates are buffered 100 at a time and flushed via `addLinksBatch` / `addTimelineEntriesBatch`; `ON CONFLICT DO NOTHING` enforces uniqueness at the DB layer, and the `created` counter returns real rows inserted (truthful on re-runs).
- `src/commands/graph-query.ts` — `gbrain graph-query <slug> [--type T] [--depth N] [--direction in|out|both]`: typed-edge relationship traversal (renders indented tree)
- `src/core/link-extraction.ts` — shared library for the v0.12.0 graph layer. extractEntityRefs (canonical, replaces backlinks.ts duplicate) matches both `[Name](people/slug)` markdown links and Obsidian `[[people/slug|Name]]` wikilinks as of v0.12.3. extractPageLinks, inferLinkType heuristics (attended/works_at/invested_in/founded/advises/source/mentions), parseTimelineEntries, isAutoLinkEnabled config helper. `DIR_PATTERN` covers `people`, `companies`, `deals`, `topics`, `concepts`, `projects`, `entities`, `tech`, `finance`, `personal`, `openclaw`. Used by extract.ts, operations.ts auto-link post-hook, and backlinks.ts.
- `src/core/minions/` — Minions job queue: BullMQ-inspired, Postgres-native (queue, worker, backoff, types, protected-names, quiet-hours, stagger, handlers/shell).
- `src/core/minions/queue.ts` — MinionQueue class (submit, claim, complete, fail, stall detection, parent-child, depth/child-cap, per-job timeouts, cascade-kill, attachments, idempotency keys, child_done inbox, removeOnComplete/Fail). `add()` takes a 4th `trusted` arg (separate from `opts` to prevent spread leakage); protected names in `PROTECTED_JOB_NAMES` require `{allowProtectedSubmit: true}` and the check runs trim-normalized (whitespace-bypass safe). v0.14.1 #219: `add()` plumbs `max_stalled` through with a `[1, 100]` clamp; omitted values let the schema DEFAULT (5) kick in.
- `src/core/minions/worker.ts` — MinionWorker class (handler registry, lock renewal, graceful shutdown, timeout safety net). v0.14.0 abort-path fix: aborted jobs now call `failJob` with reason (`timeout`/`cancel`/`lock-lost`/`shutdown`) instead of returning silently. `shutdownAbort` (instance field) fires on process SIGTERM/SIGINT and propagates to `ctx.shutdownSignal` — shell handler listens to it; non-shell handlers don't.
- `src/core/minions/types.ts` — `MinionJobInput` + `MinionJobStatus` + handler context types. `MinionJobInput.max_stalled` (new in v0.14.1) is optional; omitted values let the schema DEFAULT (5) kick in, provided values are clamped to `[1, 100]`.
- `src/core/minions/protected-names.ts` — side-effect-free constant module exporting `PROTECTED_JOB_NAMES` + `isProtectedJobName()`. Kept pure so queue core can import without loading handler modules.
- `src/core/minions/handlers/shell.ts` — `shell` job handler. Spawns `/bin/sh -c cmd` (absolute path, PATH-override-safe) or `argv[0] argv[1..]` (no shell). Env allowlist: `PATH, HOME, USER, LANG, TZ, NODE_ENV` + caller `env:` overrides. UTF-8-safe stdout/stderr tail via `string_decoder.StringDecoder`. Abort (either `ctx.signal` or `ctx.shutdownSignal`) fires SIGTERM → 5s grace → SIGKILL on child. Requires `GBRAIN_ALLOW_SHELL_JOBS=1` on worker (gated by `registerBuiltinHandlers`).
- `src/core/minions/handlers/shell-audit.ts` — per-submission JSONL audit trail at `~/.gbrain/audit/shell-jobs-YYYY-Www.jsonl` (ISO-week rotation; override via `GBRAIN_AUDIT_DIR`). Best-effort: `mkdirSync(recursive)` + `appendFileSync`; failures logged to stderr, submission not blocked. Logs cmd (first 80 chars) or argv (JSON array). Never logs env values.
- `src/core/minions/handlers/subagent.ts` (v0.15) — LLM-loop handler. Two-phase tool persistence (pending → complete/failed), replay reconciliation for mid-dispatch crashes, dual-signal abort (`ctx.signal` + `ctx.shutdownSignal`), Anthropic prompt caching on system + tool defs. `makeSubagentHandler({engine, client?, ...})` factory; `MessagesClient` is an injectable interface the real SDK implements structurally. Throws `RateLeaseUnavailableError` (renewable) when rate-lease capacity is full.
- `src/core/minions/handlers/subagent-aggregator.ts` (v0.15) — `subagent_aggregator` handler. Claims AFTER all children resolve (queue changes guarantee every terminal child posts a `child_done` inbox message with outcome). Reads inbox via `ctx.readInbox()`, builds deterministic mixed-outcome markdown summary. No LLM call in v0.15.
- `src/core/minions/handlers/subagent-audit.ts` (v0.15) — JSONL audit + heartbeat writer at `~/.gbrain/audit/subagent-jobs-YYYY-Www.jsonl`. Events: `submission` (one line per submit) + `heartbeat` (per turn boundary: `llm_call_started | llm_call_completed | tool_called | tool_result | tool_failed`). Never logs prompts or tool inputs. `readSubagentAuditForJob(jobId, {sinceIso})` is the readback path for `gbrain agent logs`.
- `src/core/minions/rate-leases.ts` (v0.15) — lease-based concurrency cap for outbound providers (default key `anthropic:messages`, max via `GBRAIN_ANTHROPIC_MAX_INFLIGHT`). Owner-tagged rows with `expires_at` auto-prune on acquire; `pg_advisory_xact_lock` guards check-then-insert; CASCADE on owning job deletion. `renewLeaseWithBackoff` retries 3x (250/500/1000ms).
- `src/core/minions/wait-for-completion.ts` (v0.15) — poll-until-terminal helper for CLI callers. `TimeoutError` does NOT cancel the job; `AbortSignal` exits without throwing. Default `pollMs`: 1000 on Postgres, 250 on PGLite inline.
- `src/core/minions/transcript.ts` (v0.15) — renders `subagent_messages` + `subagent_tool_executions` to markdown. Tool rows splice under their owning assistant `tool_use` by `tool_use_id`. UTF-8-safe truncation; unknown block types fall through to fenced JSON.
- `src/core/minions/plugin-loader.ts` (v0.15) — `GBRAIN_PLUGIN_PATH` discovery. Absolute paths only, left-wins collision, `gbrain.plugin.json` with `plugin_version: "gbrain-plugin-v1"`, plugins ship DEFS only (no new tools), `allowed_tools:` validated at load time against the derived registry.
- `src/core/minions/tools/brain-allowlist.ts` (v0.15) — derives subagent tool registry from `src/core/operations.ts`. 11-name allow-list: `query`, `search`, `get_page`, `list_pages`, `file_list`, `file_url`, `get_backlinks`, `traverse_graph`, `resolve_slugs`, `get_ingest_log`, `put_page`. `put_page` schema is namespace-wrapped per subagent (`^wiki/agents/<subagentId>/.+`); the `put_page` op's server-side check is the authoritative gate via `ctx.viaSubagent` fail-closed.
- `src/mcp/tool-defs.ts` (v0.15) — extracted `buildToolDefs(ops)` helper. MCP server + subagent tool registry both call it; byte-for-byte equivalence pinned by `test/mcp-tool-defs.test.ts`.
- `src/core/minions/attachments.ts` — Attachment validation (path traversal, null byte, oversize, base64, duplicate detection)
- `src/commands/agent.ts` (v0.16) — `gbrain agent run <prompt> [flags]` CLI. Submits `subagent` (or N children + 1 aggregator) under `{allowProtectedSubmit: true}`. Single-entry `--fanout-manifest` short-circuits. Children get `on_child_fail: 'continue'` + `max_stalled: 3`. `--follow` is the default on TTY; streams logs + polls `waitForCompletion` in parallel. Ctrl-C detaches, does not cancel.
- `src/commands/agent-logs.ts` (v0.16) — `gbrain agent logs <job> [--follow] [--since]`. Merges JSONL heartbeat audit + `subagent_messages` into a chronological timeline. `parseSince` accepts ISO-8601 or relative (`5m`, `1h`, `2d`). Transcript tail renders only for terminal jobs.
- `src/commands/jobs.ts` — `gbrain jobs` CLI subcommands + `gbrain jobs work` daemon. v0.13.1 surfaces the full `MinionJobInput` retry/backoff/timeout/idempotency surface as first-class CLI flags on `jobs submit`: `--max-stalled`, `--backoff-type fixed|exponential`, `--backoff-delay`, `--backoff-jitter`, `--timeout-ms`, `--idempotency-key`. `jobs smoke --sigkill-rescue` is the opt-in regression guard for #219. v0.16 wires `registerBuiltinHandlers` to always register `subagent` + `subagent_aggregator` (no env flag — `ANTHROPIC_API_KEY` is the natural cost gate, trust is via `PROTECTED_JOB_NAMES`) and loads `GBRAIN_PLUGIN_PATH` plugins at worker startup with a loud startup-line per plugin. `shell` handler still gated by `GBRAIN_ALLOW_SHELL_JOBS=1` (RCE surface, separate concern).
- `src/commands/features.ts` — `gbrain features --json --auto-fix`: usage scan + feature adoption salesman
- `src/commands/autopilot.ts` — `gbrain autopilot --install`: self-maintaining brain daemon (sync+extract+embed)
- `src/mcp/server.ts` — MCP stdio server (generated from operations)
- `src/commands/auth.ts` — Standalone token management (create/list/revoke/test)
- `src/commands/upgrade.ts` — Self-update CLI. `runPostUpgrade()` enumerates migrations from the TS registry (src/commands/migrations/index.ts) and tail-calls `runApplyMigrations(['--yes', '--non-interactive'])` so the mechanical side of every outstanding migration runs unconditionally.
- `src/commands/migrations/` — TS migration registry (compiled into the binary; no filesystem walk of `skills/migrations/*.md` needed at runtime). `index.ts` lists migrations in semver order. `v0_11_0.ts` = Minions adoption orchestrator (8 phases). `v0_12_0.ts` = Knowledge Graph auto-wire orchestrator (5 phases: schema → config check → backfill links → backfill timeline → verify). `phaseASchema` has a 600s timeout (bumped from 60s in v0.12.1 for duplicate-heavy brains). `v0_12_2.ts` = JSONB double-encode repair orchestrator (4 phases: schema → repair-jsonb → verify → record). `v0_14_0.ts` = shell-jobs + autopilot cooperative (2 phases: schema ALTER minion_jobs.max_stalled SET DEFAULT 3 — superseded by v0.14.3's schema-level DEFAULT 5 + UPDATE backfill; pending-host-work ping for skills/migrations/v0.14.0.md). All orchestrators are idempotent and resumable from `partial` status. As of v0.14.2 (Bug 3), the RUNNER owns all ledger writes — orchestrators return `OrchestratorResult` and `apply-migrations.ts` persists a canonical `{version, status, phases}` shape after return. Orchestrators no longer call `appendCompletedMigration` directly. `statusForVersion` prefers `complete` over `partial` (never regresses). 3 consecutive partials → wedged → `--force-retry <version>` writes a `'retry'` reset marker. v0.14.3 (fix wave) ships schema-only migrations v14 (`pages_updated_at_index`) + v15 (`minion_jobs_max_stalled_default_5` with UPDATE backfill) via the `MIGRATIONS` array in `src/core/migrate.ts` — no orchestrator phases needed.
- `src/commands/repair-jsonb.ts` — `gbrain repair-jsonb [--dry-run] [--json]`: rewrites `jsonb_typeof='string'` rows in place across 5 affected columns (pages.frontmatter, raw_data.data, ingest_log.pages_updated, files.metadata, page_versions.frontmatter). Fixes v0.12.0 double-encode bug on Postgres; PGLite no-ops. Idempotent.
- `src/commands/orphans.ts` — `gbrain orphans [--json] [--count] [--include-pseudo]`: surfaces pages with zero inbound wikilinks, grouped by domain. Auto-generated/raw/pseudo pages filtered by default. Also exposed as `find_orphans` MCP operation. Shipped in v0.12.3 (contributed by @knee5).
- `src/commands/doctor.ts` — `gbrain doctor [--json] [--fast] [--fix] [--dry-run] [--index-audit]`: health checks. v0.12.3 added `jsonb_integrity` + `markdown_body_completeness` reliability checks. v0.14.1: `--fix` delegates inlined cross-cutting rules to `> **Convention:** see [path](path).` callouts (pipes DRY violations into `src/core/dry-fix.ts`); `--fix --dry-run` previews without writing. v0.14.2: `schema_version` check fails loudly when `version=0` (migrations never ran — the #218 `bun install -g` signature) and routes users to `gbrain apply-migrations --yes`; new opt-in `--index-audit` flag (Postgres-only) reports zero-scan indexes from `pg_stat_user_indexes` (informational only, no auto-drop). v0.15.2: every DB check is wrapped in a progress phase; `markdown_body_completeness` runs under a 1s heartbeat timer so 10+ min scans are observable on 50K-page brains. Fix hints point at `gbrain repair-jsonb`, `gbrain sync --force`, and `gbrain apply-migrations`.
- `src/core/migrate.ts` — schema-migration runner. Owns the `MIGRATIONS` array (source of truth for schema DDL). v0.14.2 extended the `Migration` interface with `sqlFor?: { postgres?, pglite? }` (engine-specific SQL overrides `sql`) and `transaction?: boolean` (set to false for `CREATE INDEX CONCURRENTLY`, which Postgres refuses inside a transaction; ignored on PGLite since it has no concurrent writers). Migration v14 (fix wave) uses a handler branching on `engine.kind` to run CONCURRENTLY on Postgres (with a pre-drop of any invalid remnant via `pg_index.indisvalid`) and plain `CREATE INDEX` on PGLite. v15 bumps `minion_jobs.max_stalled` default 1→5 and backfills existing non-terminal rows.
- `src/core/progress.ts` — Shared bulk-action progress reporter. Writes to stderr. Modes: `auto` (TTY: `\r`-rewriting; non-TTY: plain lines), `human`, `json` (JSONL), `quiet`. Rate-gated by `minIntervalMs` and `minItems`. `startHeartbeat(reporter, note)` helper for single long queries. `child()` composes phase paths. Singleton SIGINT/SIGTERM coordinator emits `abort` events for every live phase. EPIPE defense on both sync throws and stream `'error'` events. Zero dependencies. Introduced in v0.15.2.
- `src/core/cli-options.ts` — Global CLI flag parser. `parseGlobalFlags(argv)` returns `{cliOpts, rest}` with `--quiet` / `--progress-json` / `--progress-interval=<ms>` stripped. `getCliOptions()` / `setCliOptions()` expose a module-level singleton so commands reach the resolved flags without parameter threading. `cliOptsToProgressOptions()` maps to reporter options. `childGlobalFlags()` returns the flag suffix to append to `execSync('gbrain ...')` calls in migration orchestrators. `OperationContext.cliOpts` extends shared-op dispatch for MCP callers.
- `src/core/cycle.ts` — v0.17 brain maintenance cycle primitive. `runCycle(engine: BrainEngine | null, opts: CycleOpts): Promise<CycleReport>` composes 6 phases in semantically-driven order (lint → backlinks → sync → extract → embed → orphans). Three callers: `gbrain dream` CLI, `gbrain autopilot` daemon's inline path, and the Minions `autopilot-cycle` handler (`src/commands/jobs.ts`). One source of truth for what the brain does overnight. Coordination via `gbrain_cycle_locks` DB table (TTL-based; works through PgBouncer transaction pooling, unlike session-scoped `pg_try_advisory_lock`) + `~/.gbrain/cycle.lock` file lock with PID-liveness for PGLite / engine=null mode. `CycleReport.schema_version: "1"` is the stable agent-consumable shape. `PhaseResult.error: { class, code, message, hint?, docs_url? }` is Stripe-API-tier structured failure info. `yieldBetweenPhases` hook awaited between every phase — Minions handler uses this to renew its job lock and prevent v0.14 stall-death regression. Engine nullable: filesystem phases (lint, backlinks) run without DB; DB phases skip with `status: "skipped", reason: "no_database"`. Lock-skip: read-only phase selections (`--phase orphans`) bypass the cycle lock.
- `src/commands/dream.ts` — v0.17 `gbrain dream` CLI. ~80-line thin alias over `runCycle`. brainDir resolution requires explicit `--dir` OR `sync.repo_path` config (no more walk-up-cwd-for-.git footgun). Flags: `--dry-run`, `--json`, `--phase <name>`, `--pull`, `--dir <path>`. Exit code 1 on status=failed (partial/warn not fatal — don't page on warnings).
- `scripts/check-progress-to-stdout.sh` — CI guard against regressing to `\r`-on-stdout progress. Wired into `bun run test` via `scripts/check-progress-to-stdout.sh && bun test` in package.json.
- `docs/progress-events.md` — Canonical JSON event schema reference. Stable from v0.15.2, additive only.
- `src/core/markdown.ts` — Frontmatter parsing + body splitter. `splitBody` requires an explicit timeline sentinel (`<!-- timeline -->`, `--- timeline ---`, or `---` immediately before `## Timeline`/`## History`). Plain `---` in body text is a markdown horizontal rule, not a separator. `inferType` auto-types `/wiki/analysis/` → analysis, `/wiki/guides/` → guide, `/wiki/hardware/` → hardware, `/wiki/architecture/` → architecture, `/writing/` → writing (plus the existing people/companies/deals/etc heuristics).
- `scripts/check-jsonb-pattern.sh` — CI grep guard. Fails the build if anyone reintroduces (a) the `${JSON.stringify(x)}::jsonb` interpolation pattern (postgres.js v3 double-encodes it), or (b) `max_stalled INTEGER NOT NULL DEFAULT 1` in any schema source file (v0.15.1 #219 regression guard — must be DEFAULT 5 to preserve SIGKILL-rescue). Wired into `bun test`.
- `scripts/llms-config.ts` + `scripts/build-llms.ts` — Generator for `llms.txt` (llmstxt.org-spec web index) + `llms-full.txt` (inlined single-fetch bundle). Curated config drives both. Run `bun run build:llms` after adding a new doc. `LLMS_REPO_BASE` env var lets forks regenerate with their own URL base. `FULL_SIZE_BUDGET` (600KB) caps the inline bundle; generator WARNs if exceeded. Committed output is not analogous to `schema-embedded.ts` (no runtime consumer); we commit for GitHub browsing and fork-safe fetching.
- `AGENTS.md` — Local-clone entry point for non-Claude agents (Codex, Cursor, OpenClaw, Aider). Mirrors `CLAUDE.md` intent via relative links. Claude Code keeps using `CLAUDE.md`.
- `docs/UPGRADING_DOWNSTREAM_AGENTS.md` — Patches for downstream agent skill forks to apply when upgrading. Each release appends a new section. v0.10.3 includes diffs for brain-ops, meeting-ingestion, signal-detector, enrich.
- `src/core/schema-embedded.ts` — AUTO-GENERATED from schema.sql (run `bun run build:schema`)
- `src/schema.sql` — Full Postgres + pgvector DDL (source of truth, generates schema-embedded.ts)
- `src/commands/integrations.ts` — Standalone integration recipe management (no DB needed). Exports `getRecipeDirs()` (trust-tagged recipe sources), SSRF helpers (`isInternalUrl`, `parseOctet`, `hostnameToOctets`, `isPrivateIpv4`). Only package-bundled recipes are `embedded=true`; `$GBRAIN_RECIPES_DIR` and cwd `./recipes/` are untrusted and cannot run `command`/`http`/string health checks.
- `src/core/search/expansion.ts` — Multi-query expansion via Haiku. Exports `sanitizeQueryForPrompt` + `sanitizeExpansionOutput` (prompt-injection defense-in-depth). Sanitized query is only used for the LLM channel; original query still drives search.
- `recipes/` — Integration recipe files (YAML frontmatter + markdown setup instructions)
- `docs/guides/` — Individual SKILLPACK guides (broken out from monolith)
- `docs/integrations/` — "Getting Data In" guides and integration docs
- `docs/architecture/infra-layer.md` — Shared infrastructure documentation
- `docs/ethos/THIN_HARNESS_FAT_SKILLS.md` — Architecture philosophy essay
- `docs/ethos/MARKDOWN_SKILLS_AS_RECIPES.md` — "Homebrew for Personal AI" essay
- `docs/guides/repo-architecture.md` — Two-repo pattern (agent vs brain)
- `docs/guides/sub-agent-routing.md` — Model routing table for sub-agents
- `docs/guides/skill-development.md` — 5-step skill development cycle + MECE
- `docs/guides/idea-capture.md` — Originality distribution, depth test, cross-linking
- `docs/guides/quiet-hours.md` — Notification hold + timezone-aware delivery
- `docs/guides/diligence-ingestion.md` — Data room to brain pages pipeline
- `docs/designs/HOMEBREW_FOR_PERSONAL_AI.md` — 10-star vision for integration system
- `docs/mcp/` — Per-client setup guides (Claude Desktop, Code, Cowork, Perplexity)
- BrainBench (benchmark suite + corpus): lives in the separate [gbrain-evals](https://github.com/garrytan/gbrain-evals) repo. Not installed alongside gbrain.
- `skills/_brain-filing-rules.md` — Cross-cutting brain filing rules (referenced by all brain-writing skills)
- `skills/RESOLVER.md` — Skill routing table (based on the agent-fork AGENTS.md pattern)
- `skills/conventions/` — Cross-cutting rules (quality, brain-first, model-routing, test-before-bulk, cross-modal)
- `skills/_output-rules.md` — Output quality standards (deterministic links, no slop, exact phrasing)
- `skills/signal-detector/SKILL.md` — Always-on idea+entity capture on every message
- `skills/brain-ops/SKILL.md` — Brain-first lookup, read-enrich-write loop, source attribution
- `skills/idea-ingest/SKILL.md` — Links/articles/tweets with author people page mandatory
- `skills/media-ingest/SKILL.md` — Video/audio/PDF/book with entity extraction
- `skills/meeting-ingestion/SKILL.md` — Transcripts with attendee enrichment chaining
- `skills/citation-fixer/SKILL.md` — Citation format auditing and fixing
- `skills/repo-architecture/SKILL.md` — Filing rules by primary subject
- `skills/skill-creator/SKILL.md` — Create conforming skills with MECE check
- `skills/daily-task-manager/SKILL.md` — Task lifecycle with priority levels
- `skills/daily-task-prep/SKILL.md` — Morning prep with calendar context
- `skills/cross-modal-review/SKILL.md` — Quality gate via second model
- `skills/cron-scheduler/SKILL.md` — Schedule staggering, quiet hours, idempotency
- `skills/reports/SKILL.md` — Timestamped reports with keyword routing
- `skills/testing/SKILL.md` — Skill validation framework
- `skills/soul-audit/SKILL.md` — 6-phase interview for SOUL.md, USER.md, ACCESS_POLICY.md, HEARTBEAT.md
- `skills/webhook-transforms/SKILL.md` — External events to brain signals
- `skills/data-research/SKILL.md` — Structured data research: email-to-tracker pipeline with parameterized YAML recipes
- `skills/minion-orchestrator/SKILL.md` — Background job orchestration: submit, fan out children with depth/cap/timeouts, collect results via child_done inbox
- `templates/` — SOUL.md, USER.md, ACCESS_POLICY.md, HEARTBEAT.md templates
- `skills/migrations/` — Version migration files with feature_pitch YAML frontmatter
- `src/commands/publish.ts` — Deterministic brain page publisher (code+skill pair, zero LLM calls)
- `src/commands/backlinks.ts` — Back-link checker and fixer (enforces Iron Law)
- `src/commands/lint.ts` — Page quality linter (catches LLM artifacts, placeholder dates)
- `src/commands/report.ts` — Structured report saver (audit trail for maintenance/enrichment)
- `openclaw.plugin.json` — ClawHub bundle plugin manifest
### BrainBench — in a sibling repo (v0.20+)
BrainBench — the public benchmark for personal-knowledge agent stacks — lives in
[github.com/garrytan/gbrain-evals](https://github.com/garrytan/gbrain-evals). It
depends on gbrain as a consumer; gbrain never pulls in the ~5MB eval corpus or
the pdf-parse dev dep at install time.
gbrain's public API surface (the exports map in `package.json`) is what
gbrain-evals consumes: `gbrain/engine`, `gbrain/types`, `gbrain/operations`,
`gbrain/pglite-engine`, `gbrain/link-extraction`, `gbrain/import-file`,
`gbrain/transcription`, `gbrain/embedding`, `gbrain/config`, `gbrain/markdown`,
`gbrain/backoff`, `gbrain/search/hybrid`, `gbrain/search/expansion`,
`gbrain/extract`. Removing any of these is a breaking change for the
gbrain-evals consumer.
## Commands
Run `gbrain --help` or `gbrain --tools-json` for full command reference.
Key commands added in v0.7:
- `gbrain init` — defaults to PGLite (no Supabase needed), scans repo size, suggests Supabase for 1000+ files
- `gbrain migrate --to supabase` / `gbrain migrate --to pglite` — bidirectional engine migration
Key commands added for Minions (job queue):
- `gbrain jobs submit <name> [--params JSON] [--follow] [--dry-run]` — submit a background job. v0.13.1 adds first-class flags for every `MinionJobInput` tuning knob: `--max-stalled N`, `--backoff-type fixed|exponential`, `--backoff-delay Nms`, `--backoff-jitter 0..1`, `--timeout-ms N`, `--idempotency-key K`.
- `gbrain jobs list [--status S] [--queue Q]` — list jobs with filters
- `gbrain jobs get <id>` — job details with attempt history
- `gbrain jobs cancel/retry/delete <id>` — manage job lifecycle
- `gbrain jobs prune [--older-than 30d]` — clean old completed/dead jobs
- `gbrain jobs stats` — job health dashboard
- `gbrain jobs smoke [--sigkill-rescue]` — health smoke test. `--sigkill-rescue` is the v0.13.1 regression guard for #219: simulates a killed worker and asserts the stalled job is requeued instead of dead-lettered on first stall.
- `gbrain jobs work [--queue Q] [--concurrency N]` — start worker daemon (Postgres only)
Key commands added in v0.12.2:
- `gbrain repair-jsonb [--dry-run] [--json]` — repair double-encoded JSONB rows left over from v0.12.0-and-earlier Postgres writes. Idempotent; PGLite no-ops. The `v0_12_2` migration runs this automatically on `gbrain upgrade`.
Key commands added in v0.12.3:
- `gbrain orphans [--json] [--count] [--include-pseudo]` — surface pages with zero inbound wikilinks, grouped by domain. Auto-generated/raw/pseudo pages filtered by default. Also exposed as `find_orphans` MCP operation. The natural consumer of the v0.12.0 knowledge graph layer: once edges are captured, find the gaps.
- `gbrain doctor` gains two new reliability detection checks: `jsonb_integrity` (v0.12.0 Postgres double-encode damage) and `markdown_body_completeness` (pages truncated by the old splitBody bug). Detection only; fix hints point at `gbrain repair-jsonb` and `gbrain sync --force`.
Key commands added in v0.14.2:
- `gbrain sync --skip-failed` — acknowledge the current set of failed-parse files recorded in `~/.gbrain/sync-failures.jsonl` so the sync bookmark advances past them. Doctor's `sync_failures` check shows previously-skipped as "all acknowledged" instead of warning.
- `gbrain sync --retry-failed` — re-walk the unacknowledged failures and re-attempt parsing. If the files now succeed, they clear from the set and the bookmark advances naturally.
- `gbrain apply-migrations --force-retry <version>` — reset a wedged migration (3 consecutive partials with no completion) by appending a `'retry'` marker. Next `apply-migrations --yes` treats the version as fresh. `complete` status never regresses to `partial` either before or after a retry marker.
- `GBRAIN_POOL_SIZE` env var — honored by both the singleton pool (`src/core/db.ts`) and the parallel-import worker pool (`src/commands/import.ts`). Default is 10; lower to 2 for Supabase transaction pooler to avoid MaxClients crashes during `gbrain upgrade` subprocess spawns. Read at call time via `resolvePoolSize()`.
- `gbrain doctor` gains two new checks: `sync_failures` (surfaces unacknowledged parse failures with exact paths + fix hints) and `brain_score` (renders the 5-component breakdown when score < 100: embed coverage / 35, link density / 25, timeline coverage / 15, orphans / 15, dead links / 10 — sum equals total).
Key commands added in v0.14.3 (fix wave):
- `gbrain doctor --index-audit` — opt-in Postgres-only check reporting zero-scan indexes from `pg_stat_user_indexes`. Informational only; never auto-drops.
- `gbrain doctor` schema_version check fails loudly when `version=0` — catches `bun install -g github:...` postinstall failures (#218) and routes users to `gbrain apply-migrations --yes`.
- `gbrain jobs submit` gains `--max-stalled`, `--backoff-type`, `--backoff-delay`, `--backoff-jitter`, `--timeout-ms`, `--idempotency-key` — exposing existing `MinionJobInput` fields as first-class CLI flags.
- `gbrain jobs smoke --sigkill-rescue` — opt-in regression smoke case simulating a killed worker; asserts the v0.14.3 schema default (`max_stalled=5`) actually rescues on first stall.
## Testing
`bun test` runs all tests. After the v0.12.1 release: ~75 unit test files + 8 E2E test files (1412 unit pass, 119 E2E when `DATABASE_URL` is set — skip gracefully otherwise). Unit tests run
without a database. E2E tests skip gracefully when `DATABASE_URL` is not set.
Unit tests: `test/markdown.test.ts` (frontmatter parsing), `test/chunkers/recursive.test.ts`
(chunking), `test/parity.test.ts` (operations contract
parity), `test/cli.test.ts` (CLI structure), `test/config.test.ts` (config redaction),
`test/files.test.ts` (MIME/hash), `test/import-file.test.ts` (import pipeline),
`test/upgrade.test.ts` (schema migrations),
`test/file-migration.test.ts` (file migration), `test/file-resolver.test.ts` (file resolution),
`test/import-resume.test.ts` (import checkpoints), `test/migrate.test.ts` (migration; v8/v9 helper-btree-index SQL structural assertions + 1000-row wall-clock fixtures that guard the O(n²)→O(n log n) fix + v0.13.1 assertions on v12/v13 SQL shape, `sqlFor` + `transaction:false` runner semantics, and the `max_stalled DEFAULT 1` regression guard),
`test/setup-branching.test.ts` (setup flow), `test/slug-validation.test.ts` (slug validation),
`test/storage.test.ts` (storage backends), `test/supabase-admin.test.ts` (Supabase admin),
`test/yaml-lite.test.ts` (YAML parsing), `test/check-update.test.ts` (version check + update CLI),
`test/pglite-engine.test.ts` (PGLite engine, all 40 BrainEngine methods including 11 cases for `addLinksBatch` / `addTimelineEntriesBatch`: empty batch, missing optionals, within-batch dedup via ON CONFLICT, missing-slug rows dropped by JOIN, half-existing batch, batch of 100 + v0.13.1 `connect()` error-wrap assertion (original error nested, #223 link in message, lock released)),
`test/engine-factory.test.ts` (engine factory + dynamic imports),
`test/integrations.test.ts` (recipe parsing, CLI routing, recipe validation),
`test/publish.test.ts` (content stripping, encryption, password generation, HTML output),
`test/backlinks.test.ts` (entity extraction, back-link detection, timeline entry generation),
`test/lint.test.ts` (LLM artifact detection, code fence stripping, frontmatter validation),
`test/report.test.ts` (report format, directory structure),
`test/skills-conformance.test.ts` (skill frontmatter + required sections validation),
`test/resolver.test.ts` (RESOLVER.md coverage, routing validation),
`test/search.test.ts` (RRF normalization, compiled truth boost, cosine similarity, dedup key),
`test/dedup.test.ts` (source-aware dedup, compiled truth guarantee, layer interactions),
`test/intent.test.ts` (query intent classification: entity/temporal/event/general),
`test/eval.test.ts` (retrieval metrics: precisionAtK, recallAtK, mrr, ndcgAtK, parseQrels),
`test/check-resolvable.test.ts` (resolver reachability, MECE overlap, gap detection, DRY checks + v0.14.1 proximity-based DRY detection + `extractDelegationTargets` coverage — 13 DRY cases),
`test/dry-fix.test.ts` (v0.14.1 auto-fix: three shape-aware expander pure-function tests, five guards — working-tree-dirty, no-git-backup, inside-code-fence, already-delegated within 40 lines, ambiguous-multi-match, block-is-callout — 28 cases),
`test/doctor-fix.test.ts` (v0.14.1 `gbrain doctor --fix` CLI integration: dry-run preview, apply path, JSON output shape — 3 cases),
`test/backoff.test.ts` (load-aware throttling, concurrency limits, active hours),
`test/fail-improve.test.ts` (deterministic/LLM cascade, JSONL logging, test generation, rotation),
`test/transcription.test.ts` (provider detection, format validation, API key errors),
`test/enrichment-service.test.ts` (entity slugification, extraction, tier escalation),
`test/data-research.test.ts` (recipe validation, MRR/ARR extraction, dedup, tracker parsing, HTML stripping),
`test/minions.test.ts` (Minions job queue v7: CRUD, state machine, backoff, stall detection, dependencies, worker lifecycle, lock management, claim mechanics, depth/child-cap, timeouts, cascade kill, idempotency, child_done inbox, attachments, removeOnComplete/Fail + v0.13.1 `max_stalled` clamp/default/plumbing coverage),
`test/extract.test.ts` (link extraction, timeline extraction, frontmatter parsing, directory type inference),
`test/extract-db.test.ts` (gbrain extract --source db: typed link inference, idempotency, --type filter, --dry-run JSON output),
`test/extract-fs.test.ts` (gbrain extract --source fs: first-run inserts + second-run reports zero, dry-run dedups candidates across files, second-run perf regression guard — the v0.12.1 N+1 dedup bug),
`test/link-extraction.test.ts` (canonical extractEntityRefs both formats, extractPageLinks dedup, inferLinkType heuristics, parseTimelineEntries date variants, isAutoLinkEnabled config),
`test/graph-query.test.ts` (direction in/out/both, type filter, indented tree output),
`test/features.test.ts` (feature scanning, brain_score calculation, CLI routing, persistence),
`test/file-upload-security.test.ts` (symlink traversal, cwd confinement, slug + filename allowlists, remote vs local trust),
`test/query-sanitization.test.ts` (prompt-injection stripping, output sanitization, structural boundary),
`test/search-limit.test.ts` (clampSearchLimit default/cap behavior across list_pages and get_ingest_log),
`test/repair-jsonb.test.ts` (v0.12.2 JSONB repair: TARGETS list, idempotency, engine-awareness),
`test/migrations-v0_12_2.test.ts` (v0.12.2 orchestrator phases: schema → repair → verify → record),
`test/markdown.test.ts` (splitBody sentinel precedence, horizontal-rule preservation, inferType wiki subtypes),
`test/orphans.test.ts` (v0.12.3 orphans command: detection, pseudo filtering, text/json/count outputs, MCP op),
`test/postgres-engine.test.ts` (v0.12.3 statement_timeout scoping: `sql.begin` + `SET LOCAL` shape, source-level grep guardrail against reintroduced bare `SET statement_timeout`),
`test/sync.test.ts` (sync logic + v0.12.3 regression guard asserting top-level `engine.transaction` is not called),
`test/doctor.test.ts` (doctor command + v0.12.3 assertions that `jsonb_integrity` scans the four v0.12.0 write sites and `markdown_body_completeness` is present),
`test/utils.test.ts` (shared SQL utilities + `tryParseEmbedding` null-return and single-warn semantics),
`test/build-llms.test.ts` (llms.txt/llms-full.txt generator: path resolution, idempotence, spec shape, regen-drift guard, content contract, AGENTS.md install-path mirror, size-budget enforcement — 7 cases),
`test/check-resolvable-cli.test.ts` (v0.19 CLI wrapper: exit codes, JSON envelope shape, AGENTS.md fallback chain),
`test/regression-v0_16_4.test.ts` (findRepoRoot regression guard — hermetic startDir parameterization),
`test/filing-audit.test.ts` (v0.19 Check 6: `writes_pages` / `writes_to` frontmatter, filing-rules JSON validation),
`test/routing-eval.test.ts` (v0.19 Check 5: fixture parsing, structural routing, ambiguous_with, Haiku tie-break layer),
`test/skill-manifest.test.ts` (v0.19 skill manifest parser: drift detection, managed-block markers),
`test/skillify-scaffold.test.ts` (v0.19 `gbrain skillify scaffold` stubs: SKILL.md, script, tests, routing-eval fixtures),
`test/skillpack-install.test.ts` (v0.19 `gbrain skillpack install` managed-block install / update / no-clobber semantics),
`test/skillpack-sync-guard.test.ts` (v0.19 sync-guard: bundled skills stay byte-identical to `skills/` source).
E2E tests (`test/e2e/`): Run against real Postgres+pgvector. Require `DATABASE_URL`.
- `bun run test:e2e` runs Tier 1 (mechanical, all operations, no API keys). Includes 9 dedicated cases for the postgres-engine `addLinksBatch` / `addTimelineEntriesBatch` bind path — postgres-js's `unnest()` binding is structurally different from PGLite's and gets its own coverage.
- `test/e2e/search-quality.test.ts` runs search quality E2E against PGLite (no API keys, in-memory)
- `test/e2e/graph-quality.test.ts` runs the v0.10.3 knowledge graph pipeline (auto-link via put_page, reconciliation, traversePaths) against PGLite in-memory
- `test/e2e/postgres-jsonb.test.ts` — v0.12.2 regression test. Round-trips all 5 JSONB write sites (pages.frontmatter, raw_data.data, ingest_log.pages_updated, files.metadata, page_versions.frontmatter) against real Postgres and asserts `jsonb_typeof='object'` plus `->>'key'` returns the expected scalar. The test that should have caught the original double-encode bug.
- `test/e2e/jsonb-roundtrip.test.ts` — v0.12.3 companion regression against the 4 doctor-scanned JSONB sites. Assertion-level overlap with `postgres-jsonb.test.ts` is intentional defense-in-depth: if doctor's scan surface ever drifts from the actual write surface, one of these tests catches it.
- `test/e2e/upgrade.test.ts` runs check-update E2E against real GitHub API (network required)
- `test/e2e/openclaw-reference-compat.test.ts` (v0.19) — exercises `check-resolvable` + `skillpack install` against a minimal AGENTS.md workspace fixture (`test/fixtures/openclaw-reference-minimal/`), regression guard for the 107-skill OpenClaw deployment shape
- Tier 2 (`skills.test.ts`) requires OpenClaw + API keys, runs nightly in CI
- If `.env.testing` doesn't exist in this directory, check sibling worktrees for one:
`find ../ -maxdepth 2 -name .env.testing -print -quit` and copy it here if found.
- Always run E2E tests when they exist. Do not skip them just because DATABASE_URL
is not set. Start the test DB, run the tests, then tear it down.
### API keys and running ALL tests
ALWAYS source the user's shell profile before running tests:
```bash
source ~/.zshrc 2>/dev/null || true
```
This loads `OPENAI_API_KEY` and `ANTHROPIC_API_KEY`. Without these, Tier 2 tests
skip silently. Do NOT skip Tier 2 tests just because they require API keys — load
the keys and run them.
When asked to "run all E2E tests" or "run tests", that means ALL tiers:
- Tier 1: `bun run test:e2e` (mechanical, sync, upgrade — no API keys needed)
- Tier 2: `test/e2e/skills.test.ts` (requires OpenAI + Anthropic + openclaw CLI)
- Always spin up the test DB, source zshrc, run everything, tear down.
### E2E test DB lifecycle (ALWAYS follow this)
You are responsible for spinning up and tearing down the test Postgres container.
Do not leave containers running after tests. Do not skip E2E tests.
1. **Check for `.env.testing`** — if missing, copy from sibling worktree.
Read it to get the DATABASE_URL (it has the port number).
2. **Check if the port is free:**
`docker ps --filter "publish=PORT"` — if another container is on that port,
pick a different port (try 5435, 5436, 5437) and start on that one instead.
3. **Start the test DB:**
```bash
docker run -d --name gbrain-test-pg \
-e POSTGRES_USER=postgres -e POSTGRES_PASSWORD=postgres \
-e POSTGRES_DB=gbrain_test \
-p PORT:5432 pgvector/pgvector:pg16
```
Wait for ready: `docker exec gbrain-test-pg pg_isready -U postgres`
4. **Run E2E tests:**
`DATABASE_URL=postgresql://postgres:postgres@localhost:PORT/gbrain_test bun run test:e2e`
5. **Tear down immediately after tests finish (pass or fail):**
`docker stop gbrain-test-pg && docker rm gbrain-test-pg`
Never leave `gbrain-test-pg` running. If you find a stale one from a previous run,
stop and remove it before starting a new one.
## Skills
Read the skill files in `skills/` before doing brain operations. GBrain ships 28 skills
organized by `skills/RESOLVER.md` (`AGENTS.md` is also accepted as of v0.19):
**Original 8 (conformance-migrated):** ingest (thin router), query, maintain, enrich,
briefing, migrate, setup, publish.
**Brain skills (ported from an upstream agent fork):** signal-detector, brain-ops, idea-ingest, media-ingest,
meeting-ingestion, citation-fixer, repo-architecture, skill-creator, daily-task-manager.
**Operational + identity:** daily-task-prep, cross-modal-review, cron-scheduler, reports,
testing, soul-audit, webhook-transforms, data-research, minion-orchestrator.
**Skillify loop (v0.19):** skillify (the markdown orchestration), skillpack-check
(agent-readable health report).
**Conventions:** `skills/conventions/` has cross-cutting rules (quality, brain-first,
model-routing, test-before-bulk, cross-modal). `skills/_brain-filing-rules.md` and
`skills/_output-rules.md` are shared references.
## Bulk-action progress reporting
All bulk commands (doctor, embed, import, export, sync, extract, migrate,
repair-jsonb, orphans, check-backlinks, lint, integrity auto, eval, files
sync, and apply-migrations) stream progress through the shared reporter
at `src/core/progress.ts`. Agents get heartbeats within 1 second of every
iteration regardless of how slow the underlying work is.
Rules:
- Progress always writes to **stderr**. Stdout stays clean for data output
(`--json` payloads, final summaries, JSON action events from `extract`).
- Non-TTY default: plain one-line-per-event human text. JSON requires the
explicit `--progress-json` flag.
- Global flags (`--quiet`, `--progress-json`, `--progress-interval=<ms>`)
are parsed by `src/core/cli-options.ts` BEFORE command dispatch.
- Phase names are machine-stable `snake_case.dot.path` (e.g.
`doctor.db_checks`, `sync.imports`). Documented in
`docs/progress-events.md`; additive changes only.
- `scripts/check-progress-to-stdout.sh` is a CI guard that fails the build
if any new code writes `\r` progress to stdout. Wired into `bun run test`.
- Minion handlers pass `job.updateProgress` as the `onProgress` callback
to core functions (DB-backed primary progress channel); stderr from
`jobs work` stays coarse for daemon liveness only.
When wiring a new bulk command: `import { createProgress } from '../core/progress.ts'`
and `import { getCliOptions, cliOptsToProgressOptions } from '../core/cli-options.ts'`.
Create a reporter with `createProgress(cliOptsToProgressOptions(getCliOptions()))`,
`start(phase, total?)` before the loop, `tick()` inside it, `finish()` after.
For single long-running queries, use `startHeartbeat(reporter, note)` with a
try/finally to guarantee cleanup. Never call `process.stdout.write('\r...')`
in bulk paths, the CI guard will fail the build.
## Build
`bun build --compile --outfile bin/gbrain src/cli.ts`
## Pre-ship requirements
Before shipping (/ship) or reviewing (/review), always run the full test suite:
- `bun test` — unit tests (no database required)
- Follow the "E2E test DB lifecycle" steps above to spin up the test DB,
run `bun run test:e2e`, then tear it down.
Both must pass. Do not ship with failing E2E tests. Do not skip E2E tests.
## Post-ship requirements (MANDATORY)
After EVERY /ship, you MUST run /document-release. This is NOT optional. Do NOT
skip it. Do NOT say "docs look fine" without running it. The skill reads every .md
file in the project, cross-references the diff, and updates anything that drifted.
If /ship's Step 8.5 triggers document-release automatically, that counts. But if
it gets skipped for ANY reason (timeout, error, oversight), you MUST run it manually
before considering the ship complete.
Files that MUST be checked on every ship:
- README.md — does it reflect new features, commands, or setup steps?
- CLAUDE.md — does it reflect new files, test files, or architecture changes?
- CHANGELOG.md — does it cover every commit?
- TODOS.md — are completed items marked done?
- docs/ — do any guides need updating?
A ship without updated docs is an incomplete ship. Period.
## CHANGELOG + VERSION are branch-scoped
**VERSION and CHANGELOG describe what THIS branch adds vs master, not how we got
here.** Every feature branch that ships gets its own version bump and CHANGELOG
entry. The entry is product release notes for users; it is not a log of internal
decisions, review rounds, or codex findings.
**Write the CHANGELOG entry at /ship time, not during development.** Mid-branch
iterations, review rounds (CEO/Eng/Codex/DX), and implementation detours belong
in the plan file at `~/.claude/plans/`, not in the CHANGELOG. One unified entry
per branch, covering what the branch added vs the base branch.
**Never edit a CHANGELOG entry that already landed on master.** If master has
v0.18.2 and your branch adds features, bump to the next version (v0.19.0, not
editing master's v0.18.2). When merging master into your branch, master may
bring new CHANGELOG entries above yours — push your entry above master's
latest and verify:
- Does CHANGELOG have your branch's own entry separate from master's entries?
- Is VERSION higher than master's VERSION?
- Is your entry the topmost `## [X.Y.Z]` entry?
- `grep "^## \[" CHANGELOG.md` shows a contiguous version sequence?
If any answer is no, fix it before continuing.
**CHANGELOG is for users, not contributors.** Write like product release notes:
- Lead with what the user can now **do** that they couldn't before. Sell the capability.
- Plain language, not implementation details. "You can now..." not "Refactored the..."
- **Never mention internal artifacts**: plan file IDs, decision tags (D-CX-#, F-ENG-#),
review rounds, codex findings, subcontractor credits. These are invisible to users.
- Put contributor-facing changes in a separate `### For contributors` section at the bottom.
- Every entry should make someone think "oh nice, I want to try that."
**What to omit:**
- "Codex caught X that the CEO review missed" — private process detail.
- "D-CX-3 split errors/warnings" — tag is meaningless to users; name the feature instead.
- "Fix-wave PR #N supersedes #M" — supersede chains belong in PR bodies, not release notes.
- "215 new cases, 3 decisions applied, 7 reviews cleared" — these are planning-mode metrics.
**What to keep:**
- The user-facing change: what commands exist now, what flag was added, what behavior fixed.
- Numbers that mean something to the user: TTHW, commands that timed out before, detection counts.
- Upgrade instructions: `gbrain upgrade` + any manual step if needed.
- Credit to external contributors when a community PR was incorporated.
## CHANGELOG voice + release-summary format
Every version entry in `CHANGELOG.md` MUST start with a release-summary section in
the GStack/Garry voice — one viewport's worth of prose + tables that lands like a
verdict, not marketing. The itemized changelog (subsections, bullets, files) goes
BELOW that summary, separated by a `### Itemized changes` header.
The release-summary section gets read by humans, by the auto-update agent, and by
anyone deciding whether to upgrade. The itemized list is for agents that need to
know exactly what changed.
### Release-summary template
Use this structure for the top of every `## [X.Y.Z]` entry:
1. **Two-line bold headline** (10-14 words total) ... should land like a verdict, not
marketing. Sound like someone who shipped today and cares whether it works.
2. **Lead paragraph** (3-5 sentences) ... what shipped, what changed for the user.
Specific, concrete, no AI vocabulary, no em dashes, no hype.
3. **A "The X numbers that matter" section** with:
- One short setup paragraph naming the source of the numbers (real production
deployment OR a reproducible benchmark ... name the file/command to run).
- A table of 3-6 key metrics with BEFORE / AFTER / Δ columns.
- A second optional table for per-category breakdown if relevant.
- 1-2 sentences interpreting the most striking number in concrete user terms.
4. **A "What this means for [audience]" closing paragraph** (2-4 sentences) tying
the metrics to a real workflow shift. End with what to do.
Voice rules:
- No em dashes (use commas, periods, "...").
- No AI vocabulary (delve, robust, comprehensive, nuanced, fundamental, etc.) or
banned phrases ("here's the kicker", "the bottom line", etc.).
- Real numbers, real file names, real commands. Not "fast" but "~30s on 30K pages."
- Short paragraphs, mix one-sentence punches with 2-3 sentence runs.
- Connect to user outcomes: "the agent does ~3x less reading" beats "improved
precision."
- Be direct about quality. "Well-designed" or "this is a mess." No dancing.
Source material to pull from:
- CHANGELOG.md previous entry for prior context
- Latest `gbrain-evals/docs/benchmarks/[latest].md` for headline numbers (sibling repo)
- Recent commits (`git log <prev-version>..HEAD --oneline`) for what shipped
- Don't make up numbers. If a metric isn't in a benchmark or production data, don't
include it. Say "no measurement yet" if asked.
Target length: ~250-350 words for the summary. Should render as one viewport.
### "To take advantage of v[version]" block (required, v0.13+)
After the release-summary and BEFORE `### Itemized changes`, every `## [X.Y.Z]`
entry MUST include a human-readable self-repair block under the heading
`## To take advantage of v[version]`.
Why: `gbrain upgrade` runs `gbrain post-upgrade` which runs `gbrain apply-migrations`.
This chain has a known weak link — `upgrade.ts` catches post-upgrade failures as
best-effort (so the binary still works). When that chain silently fails, users end
up with half-upgraded brains. The self-repair block gives them a paste-ready
recovery path; the v0.13+ `~/.gbrain/upgrade-errors.jsonl` trail + `gbrain doctor`
integration close the loop.
Template (adapt the verify commands per release):
```markdown
## To take advantage of v[version]
`gbrain upgrade` should do this automatically. If it didn't, or if `gbrain doctor`
warns about a partial migration:
1. **Run the orchestrator manually:**
```bash
gbrain apply-migrations --yes
```
2. **Your agent reads `skills/migrations/v[version].md` the next time you interact with it.**
[One sentence on whether headless agents need manual action, or whether the
orchestrator already handled the mechanical side.]
3. **Verify the outcome:**
```bash
[release-specific verify commands, e.g. `gbrain graph ... --depth 2`]
gbrain stats
```
4. **If any step fails or the numbers look wrong,** please file an issue:
https://github.com/garrytan/gbrain/issues with:
- output of `gbrain doctor`
- contents of `~/.gbrain/upgrade-errors.jsonl` if it exists
- which step broke
This feedback loop is how the gbrain maintainers find fragile upgrade paths. Thank you.
```
**Skip this block** for patches that are pure bug fixes with zero user-facing action
(rare). If the release has a schema migration, data backfill, or new feature the
user needs to verify, the block is required.
The v0.13.0 entry in CHANGELOG.md is the canonical example.
### Itemized changes (the existing rules)
Below the release summary, write `### Itemized changes` and continue with the
detailed subsections (Knowledge Graph Layer, Schema migrations, Security hardening,
Tests, etc.). Same rules as before:
- Lead with what the user can now DO that they couldn't before
- Frame as benefits and capabilities, not files changed or code written
- Make the user think "hell yeah, I want that"
- Bad: "Added GBRAIN_VERIFY.md installation verification runbook"
- Good: "Your agent now verifies the entire GBrain installation end-to-end, catching
silent sync failures and stale embeddings before they bite you"
- Bad: "Setup skill Phase H and Phase I added"
- Good: "New installs automatically set up live sync so your brain never falls behind"
- **Always credit community contributions.** When a CHANGELOG entry includes work from
a community PR, name the contributor with `Contributed by @username`. Contributors
did real work. Thank them publicly every time, no exceptions.
### Reference: v0.12.0 entry as canonical example
The v0.12.0 entry in CHANGELOG.md is the canonical example of the format. Match its
structure for every future version: bold headline, lead paragraph, "numbers that
matter" with BrainBench-style before/after table, "what this means" closer, then
`### Itemized changes` with the detailed sections below.
## Version migrations
Create a migration file at `skills/migrations/v[version].md` when a release
includes changes that existing users need to act on. The auto-update agent
reads these files post-upgrade (Section 17, Step 4) and executes them.
**You need a migration file when:**
- New setup step that existing installs don't have (e.g., v0.5.0 added live sync,
existing users need to set it up, not just new installs)
- New SKILLPACK section with a MUST ADD setup requirement
- Schema changes that require `gbrain init` or manual SQL
- Changed defaults that affect existing behavior
- Deprecated commands or flags that need replacement
- New verification steps that should run on existing installs
- New cron jobs or background processes that should be registered
**You do NOT need a migration file when:**
- Bug fixes with no behavior changes
- Documentation-only improvements (the agent re-reads docs automatically)
- New optional features that don't affect existing setups
- Performance improvements that are transparent
**The key test:** if an existing user upgrades and does nothing else, will their
brain work worse than before? If yes, migration file. If no, skip it.
Write migration files as agent instructions, not technical notes. Tell the agent
what to do, step by step, with exact commands. See `skills/migrations/v0.5.0.md`
for the pattern.
## Migration is canonical, not advisory
GBrain's job is to deliver a canonical, working setup to every user on upgrade.
Anything that looks like a "host-repo change" — AGENTS.md, cron manifests,
launchctl units, config files outside `~/.gbrain/` — is a GBrain migration
step, not a nudge we leave for the host-repo maintainer. Migrations edit host
files (with backups) to make the canonical setup real. Exceptions: changes
that require human judgment (content edits, renames that break semantics,
host-specific handler registration where shell-exec would be an RCE surface).
Everything mechanical ships in the migration.
**Test:** if shipping a feature requires a sentence that starts with "in
your AGENTS.md, add…" or "in your cron/jobs.json, rewrite…", the migration
orchestrator should be doing that edit, not the user.
**The exception is host-specific code.** For custom Minion handlers
(host-specific integrations like inbox sweeps or third-party API scanners), shipping them as a
data file the worker would exec is an RCE surface. Those get registered in
the host's own repo via the plugin contract (`docs/guides/plugin-handlers.md`);
the migration orchestrator emits a structured TODO to
`~/.gbrain/migrations/pending-host-work.jsonl` + the host agent walks the
TODOs using `skills/migrations/v0.11.0.md` — stays host-agnostic, still
canonical.
## Privacy rule: scrub real names from public docs
**Never reference real people, companies, funds, or private agent names in any
public-facing artifact.** Public artifacts include: `CHANGELOG.md`, `README.md`,
`docs/`, `skills/`, PR titles + bodies, commit messages, and comments in checked-in
code. Query examples, benchmark stories, and migration guides MUST use generic
placeholders.
Why: gbrain runs a personal knowledge brain containing notes on real people and
real companies (YC founders, portfolio companies, funds, investors, meeting
attendees). When a doc copies a query like `gbrain graph diana-hu --depth 2` or
names a specific agent fork like `Wintermute`, that real name gets indexed by
search engines, surfaced in cross-references, and distributed with every release.
**Name mapping** to use in examples:
- Agent forks → `your agent fork`, `a downstream agent`, or `agent-fork`
- Example person → `alice-example`, `charlie-example`, or `a-founder`
- Example company → `acme-example`, `widget-co`, or `a-company`
- Example fund → `fund-a`, `fund-b`, `fund-c`
- Example deal → `acme-seed`, `widget-series-a`
- Example meeting → `meetings/2026-04-03` (generic date is fine)
- Example user → `you` or `the user`, never a proper name
**Specific rule: never say `Wintermute` in any CHANGELOG, README, doc, PR, or
commit message.** When the temptation is to illustrate with the real fork name:
- Reader-facing copy → `your OpenClaw` (covers Wintermute, Hermes, AlphaClaw,
and any other downstream OpenClaw deployment in one term the reader already
recognizes).
- First-person / origin-story copy → `Garry's OpenClaw` (honest that this is
the production deployment driving the feature, without exposing the private
agent's name).
`Wintermute` may appear in private artifacts (scratch plans under
`~/.gstack/projects/…`, memory files, conversation transcripts, CEO-review
plans) — those aren't distributed. Anything checked into this repo or shipped
in a release must use the OpenClaw phrasing above. Sweeping a stale reference
is a small clean-up PR, not a debate.
**When in doubt, ask yourself:** "Would this query reveal private information
about the user's contacts, investments, or portfolio if it were read by a
stranger?" If yes, replace with generic placeholders.
**Illustrative API examples with household-brand companies** (Stripe, Brex, OpenAI,
GitHub, etc.) are fine — they're public entities, not contacts in anyone's brain.
Do not confuse illustrative API examples with queries that reveal real
relationships.
## Responsible-disclosure rule: don't broadcast attack surface in release notes
**When a release fixes a security gap or a user-impacting bug, describe the fix
functionally. Do not enumerate the attack surface, quantify the exposure window,
or highlight the most sensitive records by name in public-facing artifacts.**
Public-facing artifacts include: `CHANGELOG.md`, `README.md`, `docs/`, PR titles
and bodies, commit messages, GitHub issue titles and comments, release pages,
tweets, blog posts.
**Don't write:**
- "10 tables were publicly readable by the anon key for months, including X, Y, Z"
- "X and Y are the most sensitive ones"
- "N tables exposed. Fix: enable RLS on these specific tables: ..."
**Do write:**
- "Security hardening pass. Fresh installs secure by default. Existing brains
brought to the same bar automatically on upgrade."
- "If `gbrain doctor` still flags anything after upgrade, the message names each
table and gives the exact fix."
Why: anyone reading the release page before they've upgraded now has a directed
probe list for unpatched installs. The source code ships the specifics anyway
(`src/schema.sql`, `src/core/migrate.ts`, test fixtures) — reverse engineers can
get them. But the release page is a broadcast channel. Don't hand attackers a
curated list with a banner.
**The test:** if a reader with no prior context could read the release note and
walk away knowing "gbrain at version X has table Y readable by anon key until
they patch," the note is too specific. Rewrite until that's no longer possible.
**What IS fine in public artifacts:**
- The mechanism of the fix ("the check now scans every public table instead of
a hardcoded allowlist").
- User-facing operator ergonomics (the escape-hatch SQL template, the upgrade
commands, the breaking-change flag).
- Credit to contributors.
- Generic framing of severity ("security posture tightening pass") without
quantification.
**What stays in private artifacts (plan files, private memories, internal docs):**
- Specific table names, record counts, exposure duration.
- Which records stand out as highest-risk.
- Detailed before/after tables in the "numbers that matter" format.
If the CEO/Eng review of a plan produces a detailed exposure table, keep it in
the plan file under `~/.claude/plans/` or `~/.gstack/projects/`. Don't copy it
into the CHANGELOG or PR body.
Applies retroactively: if you see a prior CHANGELOG entry naming attack-surface
specifics, scrub it as a small cleanup commit, the same way a stale Wintermute
reference gets swept.
## Schema state tracking
`~/.gbrain/update-state.json` tracks which recommended schema directories the user
adopted, declined, or added custom. The auto-update agent (SKILLPACK Section 17)
reads this during upgrades to suggest new schema additions without re-suggesting
things the user already declined. The setup skill writes the initial state during
Phase C/E. Never modify a user's custom directories or re-suggest declined ones.
## GitHub Actions SHA maintenance
All GitHub Actions in `.github/workflows/` are pinned to commit SHAs. Before shipping
(`/ship`) or reviewing (`/review`), check for stale pins and update them:
```bash
for action in actions/checkout oven-sh/setup-bun actions/upload-artifact actions/download-artifact softprops/action-gh-release gitleaks/gitleaks-action; do
tag=$(grep -r "$action@" .github/workflows/ | head -1 | grep -o '#.*' | tr -d '# ')
[ -n "$tag" ] && echo "$action@$tag: $(gh api repos/$action/git/ref/tags/$tag --jq .object.sha 2>/dev/null)"
done
```
If any SHA differs from what's in the workflow files, update the pin and version comment.
## PR descriptions cover the whole branch
Pull request titles and bodies must describe **everything in the PR diff against the
base branch**, not just the most recent commit you made. When you open or update a
PR, walk the full commit range with `git log --oneline <base>..<head>` and write the
body to cover all of it. Group by feature area (schema, code, tests, docs) — not
chronologically by commit.
This matters because reviewers read the PR body to understand what's shipping. If
the body only covers your last commit, they miss everything else and can't review
properly. A 7-commit PR with a body that describes commit 7 is worse than no body
at all — it actively misleads.
When in doubt, run `gh pr view <N> --json commits --jq '[.commits[].messageHeadline]'`
to see what's actually in the PR before writing the body.
## Community PR wave process
Never merge external PRs directly into master. Instead, use the "fix wave" workflow:
1. **Categorize** — group PRs by theme (bug fixes, features, infra, docs)
2. **Deduplicate** — if two PRs fix the same thing, pick the one that changes fewer
lines. Close the other with a note pointing to the winner.
3. **Collector branch** — create a feature branch (e.g. `garrytan/fix-wave-N`), cherry-pick
or manually re-implement the best fixes from each PR. Do NOT merge PR branches directly —
read the diff, understand the fix, and write it yourself if needed.
4. **Test the wave** — verify with `bun test && bun run test:e2e` (full E2E lifecycle).
Every fix in the wave must have test coverage.
5. **Close with context** — every closed PR gets a comment explaining why and what (if
anything) supersedes it. Contributors did real work; respect that with clear communication
and thank them.
6. **Ship as one PR** — single PR to master with all attributions preserved via
`Co-Authored-By:` trailers. Include a summary of what merged and what closed.
**Community PR guardrails:**
- Always AskUserQuestion before accepting commits that touch voice, tone, or
promotional material (README intro, CHANGELOG voice, skill templates).
- Never auto-merge PRs that remove YC references or "neutralize" the founder perspective.
- Preserve contributor attribution in commit messages.
## Skill routing
When the user's request matches an available skill, ALWAYS invoke it using the Skill
tool as your FIRST action. Do NOT answer directly, do NOT use other tools first.
The skill has specialized workflows that produce better results than ad-hoc answers.
**NEVER hand-roll ship operations.** Do not manually run git commit + push + gh pr
create when /ship is available. /ship handles VERSION bump, CHANGELOG, document-release,
pre-landing review, test coverage audit, and adversarial review. Manually creating a PR
skips all of these. If the user says "commit and ship", "push and ship", "bisect and
ship", or any combination that ends with shipping — invoke /ship and let it handle
everything including the commits. If the branch name contains a version (e.g.
`v0.5-live-sync`), /ship should use that version for the bump.
Key routing rules:
- Product ideas, "is this worth building", brainstorming → invoke office-hours
- Bugs, errors, "why is this broken", 500 errors → invoke investigate
- Ship, deploy, push, create PR, "commit and ship", "push and ship" → invoke ship
- QA, test the site, find bugs → invoke qa
- Code review, check my diff → invoke review
- Update docs after shipping → invoke document-release
- Weekly retro → invoke retro
- Design system, brand → invoke design-consultation
- Visual audit, design polish → invoke design-review
- Architecture review → invoke plan-eng-review
- Save progress, checkpoint, resume → invoke checkpoint
- Code quality, health check → invoke health
Reference in New Issue View Git Blame Copy Permalink