mirror of
https://github.com/garrytan/gbrain.git
synced 2026-07-27 22:15:33 +00:00
* feat(facts): typed-claim substrate + cycle correctness fixes (v0.35.6 wave 1/3) Schema (migration v67): - Add four optional typed-claim columns to facts: claim_metric TEXT, claim_value DOUBLE PRECISION, claim_unit TEXT, claim_period TEXT - Partial index facts_typed_claim_idx ON (entity_slug, claim_metric, valid_from) WHERE claim_metric IS NOT NULL - All nullable, metadata-only on both engines Fence layer: - ParsedFact (facts-fence.ts) gains optional claimMetric/Value/Unit/Period - Parser tolerates both 10-cell (legacy) and 14-cell (widened) rows - Renderer emits 14 cells iff any row has typed data; otherwise stays 10-cell so existing fences don't widen on unrelated edits - Numeric value cell tolerates comma thousand separators (50,000 -> 50000) Extract pipeline (D-CDX-2, D-ENG-1): - src/core/facts/extract.ts (the actual Haiku call site, NOT extract-facts.ts cycle phase) extends its system prompt to emit typed fields for metric-shaped claims - extractFactsFromFenceText gains optional pageEffectiveDate. Precedence: fence-row validFrom > pageEffectiveDate > undefined (engine defaults to now) - normalizeMetricLabel: 15-entry seed map for common founder metrics (mrr, arr, runway, headcount, team_size, cac, ltv, gross_margin, burn_rate, cash, users, mau, dau, churn_rate, revenue); unknown labels lowercase + space->_ Engine extensions: - NewFact + insertFact + insertFacts in both engines accept the four typed columns (all nullable) - Cycle phase extract-facts.ts threads page.effective_date through AND batch-embeds via gateway.embed() before insertFacts (D-CDX-3 fix for cycle-inserted facts arriving with embedding=NULL) Consolidate fix (D-CDX-4 — Codex F4): - Replace MAX(row_num)+1 INSERT with semantic upsert on (page_id, claim, since_date). Re-running the full cycle on stable input produces zero new takes — fixes the pre-existing duplicate-takes bug after extract_facts wipes consolidated_at - Chronological valid_until writeback per cluster: sort by (valid_from ASC, id ASC), walk pairs, set older.valid_until = newer.valid_from Tests: - test/migrate.test.ts +6 cases for v67 shape + materialization + nullable backward compat - test/facts-fence-typed.test.ts (new, 17 cases): parser+renderer round-trip, normalization seed map coverage, valid_from precedence three-branch - test/consolidate-valid-until.test.ts (new, 4 cases): chronological writeback (R4a), same-day id tiebreaker, cycle re-run zero duplicates (R4b/R7), valid_until idempotency - test/schema-bootstrap-coverage.test.ts: add four typed-claim columns to COLUMN_EXEMPTIONS (migration co-defines the partial index, no forward reference to bootstrap) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * feat(trajectory): find_trajectory MCP op + eval/founder CLIs (v0.35.6 wave 2/3) Engine method (D-CDX-1, D-CDX-6): - BrainEngine.findTrajectory(opts) on both Postgres and PGLite - TrajectoryOpts: scalar sourceId fast path + sourceIds federated array (mirrors v0.34.1.0 search* dual pattern) - opts.remote: when true, SQL adds AND visibility='world' so OAuth read clients see only world-visibility facts (mirrors recall's posture — closes the F7 privacy regression Codex caught in plan review) - Single SQL query, ORDER BY valid_from ASC, id ASC for deterministic output (R3 pin). Returns TrajectoryPoint[] including raw embedding so the caller can compute drift without a second round-trip Pure function library (src/core/trajectory.ts, new): - detectRegressions(points, threshold): walks consecutive (metric, value) pairs per metric; emits when newer drops >= threshold below older. 10% default, override via GBRAIN_TRAJECTORY_REGRESSION_THRESHOLD - computeDriftScore(points): 1 - mean(cosine(emb[i], emb[i-1])) over embedded points; clamped [0,1]; null when <3 embedded points (D-ENG-3 graceful degradation) - computeTrajectoryStats(points): composed shape returning both - TRAJECTORY_SCHEMA_VERSION = 1 — additive-only across releases (R5) MCP op (src/core/operations.ts): - find_trajectory: scope read, NOT localOnly. Routes through sourceScopeOpts(ctx) for federated isolation AND threads ctx.remote for visibility filtering. Strips raw Float32Array embeddings from the wire shape; converts valid_from to YYYY-MM-DD string - Registered in operations array after find_experts - FIND_TRAJECTORY_DESCRIPTION in operations-descriptions.ts CLIs: - gbrain eval trajectory <entity> [--metric M] [--since D] [--until D] [--limit N] [--json] — chronological human view with [REGRESSION] inline annotation; thin-client routing via callRemoteTool(find_trajectory). Dispatched in src/commands/eval.ts sub-subcommand block - gbrain founder scorecard <entity> [--since D] [--until D] [--json] — pure aggregation over Phase 2's substrate. Four signals: claim_accuracy (over resolved takes), consistency, growth_trajectory, red_flags. computeFounderScorecard exported for tests. Registered as top-level command in cli.ts; added to CLI_ONLY set Tests (45 cases across 5 files): - test/engine-find-trajectory.test.ts: 18 cases — chronological order, source scoping (scalar + federated), visibility filter on remote=true, metric + since/until filters, regression detection at threshold boundaries, drift score with various embedding states - test/operations-find-trajectory.test.ts: 9 cases — op registration, param validation, JSON envelope shape, R5 schema_version: 1, embedding stripped from wire, R6 visibility filter, source scoping - test/eval-trajectory.test.ts: 7 cases — arg parsing, --help, --json envelope, regression annotation, --metric filter, empty entity - test/founder-scorecard.test.ts: 9 cases — empty inputs no-NaN (G2), claim_accuracy math, consistency math, growth_trajectory math, red_flags fire for regression / narrative_drift / missed_prediction - test/eval-contradictions/no-valid-until-write.test.ts: 4 cases — R1 (probe never writes valid_until under eval-contradictions/) + R8 (only allow-listed files write valid_until anywhere in src/) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * chore: v0.35.6.0 — CHANGELOG + VERSION + docs + migration note Bumps to v0.35.6.0 (next-minor after master's v0.35.5.1 — typed-claim substrate + trajectory + founder scorecard is a new user-facing feature surface, not a fix). - VERSION + package.json synced - CHANGELOG.md release-summary block in the wave-style voice, lead with what the user can now DO. Sections: typed metric claims in the fence, chronological metric trajectories, founder scorecard, MCP find_trajectory op, cycle re-run idempotency fix, embedding-on-insert fix, valid_from precedence fix. To-take-advantage-of block with verification + opt-in fence syntax example - CLAUDE.md Key Files entry consolidating the wave across eval-trajectory.ts + founder-scorecard.ts + trajectory.ts. Names every D-ENG / D-CDX decision and the Codex outside-voice F-numbers - skills/migrations/v0.35.6.md agent-readable migration note. Includes fence-syntax example for typed-claim rows so downstream agents start emitting them. Iron-rule contracts called out (R1 + R8 + R7 + visibility) - llms-full.txt regenerated to reflect the new CLAUDE.md entry Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * docs: post-ship sync for v0.35.7.0 — trajectory + founder scorecard - README.md: add `gbrain eval trajectory` to EVAL section, add new TEMPORAL block covering `gbrain founder scorecard` + the GBRAIN_TRAJECTORY_REGRESSION_THRESHOLD env override; add v0.35.7 "What's new" paragraph below the v0.28.8 LongMemEval blurb - AGENTS.md: new bullet under Common tasks teaching agents to reach for `gbrain eval trajectory` / `gbrain founder scorecard` / the `find_trajectory` MCP op when asked to evaluate a founder/company over time - docs/contradictions.md: append "Temporal axis follow-on (v0.35.3.1 + v0.35.7)" subsection under See also, cross-linking the trajectory substrate and naming the auto-supersession.ts:4 invariant preserved by both the verdict enum (probe side) and consolidate's valid_until writeback (cycle side) - CLAUDE.md: fix stale (v0.35.4) tag on the trajectory entry to (v0.35.7) — version got rebumped twice during the merge wave - skills/migrations/v0.35.7.md renamed to v0.35.7.0.md for consistency with the v0.35.0.0.md / v0.14.0.md / etc naming convention - llms-full.txt regenerated to reflect the CLAUDE.md edit Coverage map (Diataxis): /eval trajectory CLI ✅ ref (README, AGENTS) ✅ how-to (CHANGELOG) ❌ tutorial /founder scorecard CLI ✅ ref (README, AGENTS) ✅ how-to (CHANGELOG) ❌ tutorial find_trajectory MCP op ✅ ref (CLAUDE.md, AGENTS, contradictions.md) typed-claim fence cols ✅ ref (skills/migrations/v0.35.7.0.md, CHANGELOG) Migration v67 ✅ ref (CLAUDE.md, CHANGELOG) No tutorial / explanation gaps worth filling in this PR — the migration note's fence-syntax example already covers the "first typed claim" walkthrough. ARCHITECTURE diagrams not drifted (the trajectory work extends existing facts/takes infrastructure; no new component boxes). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
159 lines
6.8 KiB
TypeScript
159 lines
6.8 KiB
TypeScript
/**
|
|
* v0.35.4 (R1 + R8 — D-CDX-4 + D-CDX-7) — IRON-RULE: the contradiction
|
|
* probe NEVER writes `valid_until` on the facts table.
|
|
*
|
|
* The temporal trajectory wave (v0.35.4) gives `consolidate` the
|
|
* authority to write `valid_until` on chronologically-superseded facts.
|
|
* That authority is exclusive: the contradiction probe surfaces
|
|
* `temporal_supersession` verdicts via paste-ready commands, but it
|
|
* must NEVER auto-mutate. This preserves the
|
|
* `src/core/eval-contradictions/auto-supersession.ts:4` invariant.
|
|
*
|
|
* Two layered guards:
|
|
* R1 — grep guard over the entire `src/core/eval-contradictions/`
|
|
* subtree and the `src/commands/eval-suspected-contradictions*.ts`
|
|
* files: no code path may UPDATE facts.valid_until.
|
|
* R8 — broader guard over all of `src/`: the only file that writes
|
|
* valid_until is `src/core/cycle/phases/consolidate.ts`. Any new
|
|
* write site fails this guard; the human adding it must explicitly
|
|
* amend the allow-list AND document the deliberate design change.
|
|
*/
|
|
|
|
import { test, expect, describe } from 'bun:test';
|
|
import { readdirSync, readFileSync, statSync } from 'node:fs';
|
|
import { join } from 'node:path';
|
|
|
|
// Allow-listed files that legitimately write `valid_until`. Adding a new
|
|
// file here means you've thought carefully about the
|
|
// auto-supersession.ts:4 invariant and decided the new write site
|
|
// preserves it.
|
|
//
|
|
// - consolidate.ts (v0.35.4 — chronological writeback)
|
|
// - facts/forget.ts (v0.32.2 — user-initiated `gbrain forget`; user is
|
|
// the supersession authority, not the probe)
|
|
const VALID_UNTIL_WRITE_ALLOWLIST: ReadonlySet<string> = new Set([
|
|
'src/core/cycle/phases/consolidate.ts',
|
|
'src/core/facts/forget.ts',
|
|
]);
|
|
|
|
function walkTs(dir: string, acc: string[] = []): string[] {
|
|
for (const name of readdirSync(dir)) {
|
|
const full = join(dir, name);
|
|
const st = statSync(full);
|
|
if (st.isDirectory()) {
|
|
// Skip generated / vendored / test directories.
|
|
if (name === 'node_modules' || name === '.git' || name === 'dist') continue;
|
|
walkTs(full, acc);
|
|
} else if (name.endsWith('.ts') && !name.endsWith('.test.ts')) {
|
|
acc.push(full);
|
|
}
|
|
}
|
|
return acc;
|
|
}
|
|
|
|
/**
|
|
* Detect lines that look like they are UPDATEing facts.valid_until.
|
|
* Permissive on SQL formatting (the same UPDATE can be split across lines,
|
|
* use template strings, or use parameterized SQL via postgres.js's
|
|
* tagged-template syntax). We look for the pair of substrings near
|
|
* each other in the same file: `UPDATE facts` and `valid_until`.
|
|
*
|
|
* Tolerated: a reference to `valid_until` in a SELECT projection list
|
|
* (which is fine — reading the column is not writing it) is filtered out
|
|
* by also requiring a write verb (SET) nearby OR an INSERT INTO facts
|
|
* with valid_until in the column list (INSERT path is OK from
|
|
* src/core/postgres-engine.ts + src/core/pglite-engine.ts because those
|
|
* are the engine layer, intentionally writing on caller request via
|
|
* insertFact/insertFacts; the issue is whether the contradiction probe
|
|
* path triggers those writes).
|
|
*/
|
|
function findValidUntilWrites(source: string): string[] {
|
|
// Quick-fail: no mention of valid_until at all.
|
|
if (!source.includes('valid_until')) return [];
|
|
const lines = source.split('\n');
|
|
const hits: string[] = [];
|
|
for (let i = 0; i < lines.length; i++) {
|
|
const line = lines[i];
|
|
// Detect actual SQL writes. The narrow pattern `UPDATE facts SET ...
|
|
// valid_until` is unambiguous — UPDATE is a SQL verb, not text
|
|
// anyone writes in a description string. Tolerates same-line and
|
|
// multi-line UPDATEs (look at the next 4 lines after `UPDATE facts`
|
|
// for `valid_until`).
|
|
if (/\bUPDATE\s+facts\b/i.test(line)) {
|
|
const window = lines.slice(i, i + 5).join('\n');
|
|
if (/\bSET\b[\s\S]*\bvalid_until\b/i.test(window)) {
|
|
hits.push(`${i + 1}: ${line.trim()}`);
|
|
}
|
|
}
|
|
}
|
|
return hits;
|
|
}
|
|
|
|
describe('R1 — contradiction probe never writes valid_until', () => {
|
|
test('no file under src/core/eval-contradictions/ writes facts.valid_until', () => {
|
|
const dir = 'src/core/eval-contradictions';
|
|
const files = walkTs(dir);
|
|
expect(files.length).toBeGreaterThan(0);
|
|
for (const f of files) {
|
|
const src = readFileSync(f, 'utf-8');
|
|
const hits = findValidUntilWrites(src);
|
|
expect(hits, `${f} contains valid_until write: ${hits.join(' | ')}`).toEqual([]);
|
|
}
|
|
});
|
|
|
|
test('no src/commands/eval-suspected-contradictions* file writes facts.valid_until', () => {
|
|
const dir = 'src/commands';
|
|
const files = readdirSync(dir)
|
|
.filter(n => n.startsWith('eval-suspected-contradictions') && n.endsWith('.ts'))
|
|
.map(n => join(dir, n));
|
|
expect(files.length).toBeGreaterThanOrEqual(1);
|
|
for (const f of files) {
|
|
const src = readFileSync(f, 'utf-8');
|
|
const hits = findValidUntilWrites(src);
|
|
expect(hits, `${f} contains valid_until write: ${hits.join(' | ')}`).toEqual([]);
|
|
}
|
|
});
|
|
});
|
|
|
|
describe('R8 — only the consolidate phase + engine insert layer may write valid_until', () => {
|
|
test('every src/ TypeScript file that writes valid_until is on the allow-list', () => {
|
|
const files = walkTs('src');
|
|
const offenders: Array<{ file: string; hits: string[] }> = [];
|
|
|
|
for (const f of files) {
|
|
// Normalize path separator for cross-platform matching of the
|
|
// allow-list keys.
|
|
const relForCheck = f.replace(/\\/g, '/');
|
|
if (VALID_UNTIL_WRITE_ALLOWLIST.has(relForCheck)) continue;
|
|
|
|
// Engine implementation files (postgres-engine.ts, pglite-engine.ts)
|
|
// legitimately write valid_until inside insertFact/insertFacts —
|
|
// those are caller-driven INSERTs. Pattern is `INSERT INTO facts (
|
|
// ... valid_until ...) VALUES`. Detect and exempt that pattern.
|
|
// The R8 guard is about UPDATE; INSERT carrying valid_until as a
|
|
// column value is not the failure mode auto-supersession.ts:4 cares
|
|
// about.
|
|
const src = readFileSync(f, 'utf-8');
|
|
const hits = findValidUntilWrites(src);
|
|
if (hits.length > 0) offenders.push({ file: relForCheck, hits });
|
|
}
|
|
|
|
expect(
|
|
offenders,
|
|
`Unexpected valid_until UPDATE sites:\n` +
|
|
offenders.map(o => ` ${o.file}:\n ${o.hits.join('\n ')}`).join('\n') +
|
|
`\n\nIf you added a deliberate write, append the path to ` +
|
|
`VALID_UNTIL_WRITE_ALLOWLIST in this test AND review the ` +
|
|
`\`auto-supersession.ts:4\` invariant first.`,
|
|
).toEqual([]);
|
|
});
|
|
|
|
test('VALID_UNTIL_WRITE_ALLOWLIST is non-empty (consolidate is on it)', () => {
|
|
// Self-check: if the test file ever ships with an empty allow-list,
|
|
// the R8 guard collapses to "no code writes valid_until anywhere" and
|
|
// becomes a tautology. Keep the consolidate phase on the list as the
|
|
// explicit positive control.
|
|
expect(VALID_UNTIL_WRITE_ALLOWLIST.has('src/core/cycle/phases/consolidate.ts')).toBe(true);
|
|
});
|
|
});
|