Files
gbrain/test/consolidate-valid-until.test.ts
T
1dadd9ed71 v0.35.7.0 feat: temporal trajectory + founder scorecard (Phases 2-4) (#1131)
* 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>
2026-05-17 18:52:38 -07:00

227 lines
9.4 KiB
TypeScript

/**
* v0.35.4 (D-CDX-4) — consolidate semantic upsert + chronological
* valid_until writeback.
*
* Pins:
* - R4a: a cluster of 3 chronologically-ordered facts produces
* 2 facts with valid_until set (older) and 1 with NULL (newest).
* - R4b/R7: running consolidate twice on the same input produces zero
* NEW takes (semantic upsert by (page_id, claim, since_date)).
* This is the Codex F4 fix — without it, the second cycle's
* extract_facts would clear consolidated_at and the second
* consolidate would append duplicate takes via MAX(row_num)+1.
*/
import { describe, test, expect, beforeAll, afterAll, beforeEach } from 'bun:test';
import { PGLiteEngine } from '../src/core/pglite-engine.ts';
import { runPhaseConsolidate } from '../src/core/cycle/phases/consolidate.ts';
let engine: PGLiteEngine;
beforeAll(async () => {
engine = new PGLiteEngine();
await engine.connect({});
await engine.initSchema();
});
afterAll(async () => {
await engine.disconnect();
});
beforeEach(async () => {
await engine.executeRaw(`DELETE FROM facts`);
await engine.executeRaw(`DELETE FROM takes`);
await engine.executeRaw(`DELETE FROM pages WHERE slug LIKE 'cdx4-%'`);
});
function unitVec(): string {
const a = new Float32Array(1536);
a[0] = 1.0;
return '[' + Array.from(a).join(',') + ']';
}
async function seedPage(slug: string): Promise<number> {
await engine.executeRaw(
`INSERT INTO pages (slug, type, title) VALUES ($1, 'company', 'Test') ON CONFLICT DO NOTHING`,
[slug],
);
const r = await engine.executeRaw<{ id: number }>(
`SELECT id FROM pages WHERE slug = $1 AND source_id = 'default'`,
[slug],
);
return r[0].id;
}
async function insertFact(args: {
entity_slug: string;
text: string;
valid_from: Date;
confidence?: number;
}): Promise<number> {
const r = await engine.executeRaw<{ id: number }>(
`INSERT INTO facts (source_id, entity_slug, fact, kind, source, valid_from, confidence, embedding, embedded_at)
VALUES ('default', $1, $2, 'fact', 'test', $3::timestamptz, $4, $5::vector, $3::timestamptz)
RETURNING id`,
[args.entity_slug, args.text, args.valid_from.toISOString(), args.confidence ?? 0.9, unitVec()],
);
return r[0].id;
}
describe('R4a — chronological valid_until writeback', () => {
test('cluster of 3 chronologically-ordered facts: 2 older get valid_until set, newest stays NULL', async () => {
await seedPage('cdx4-acme-mrr');
const olderDay = new Date('2026-01-15T00:00:00Z');
const midDay = new Date('2026-04-12T00:00:00Z');
const newest = new Date('2026-07-08T00:00:00Z');
// All three close enough in vector space to cluster together (identical
// embeddings via unitVec()). Past the 24h "oldest age" gate.
const idOlder = await insertFact({
entity_slug: 'cdx4-acme-mrr',
text: 'MRR claim',
valid_from: olderDay,
});
const idMid = await insertFact({
entity_slug: 'cdx4-acme-mrr',
text: 'MRR claim',
valid_from: midDay,
});
const idNewest = await insertFact({
entity_slug: 'cdx4-acme-mrr',
text: 'MRR claim',
valid_from: newest,
});
const r = await runPhaseConsolidate(engine, {});
expect(r.details.facts_consolidated).toBe(3);
expect(r.details.takes_written).toBe(1);
const rows = await engine.executeRaw<{ id: number; valid_until: Date | null }>(
`SELECT id, valid_until FROM facts WHERE entity_slug = 'cdx4-acme-mrr' ORDER BY valid_from ASC`,
);
expect(rows.length).toBe(3);
// Older fact's valid_until = mid.valid_from.
expect(rows[0].id).toBe(idOlder);
expect(rows[0].valid_until).not.toBeNull();
expect(new Date(rows[0].valid_until!).toISOString().slice(0, 10)).toBe('2026-04-12');
// Mid fact's valid_until = newest.valid_from.
expect(rows[1].id).toBe(idMid);
expect(rows[1].valid_until).not.toBeNull();
expect(new Date(rows[1].valid_until!).toISOString().slice(0, 10)).toBe('2026-07-08');
// Newest fact's valid_until stays NULL.
expect(rows[2].id).toBe(idNewest);
expect(rows[2].valid_until).toBeNull();
});
test('same-day cluster (3 facts, identical valid_from): id tiebreaker establishes chronological order', async () => {
await seedPage('cdx4-acme-sameday');
const sameDay = new Date(Date.now() - 30 * 60 * 60 * 1000);
const idA = await insertFact({ entity_slug: 'cdx4-acme-sameday', text: 'same day', valid_from: sameDay });
const idB = await insertFact({ entity_slug: 'cdx4-acme-sameday', text: 'same day', valid_from: sameDay });
const idC = await insertFact({ entity_slug: 'cdx4-acme-sameday', text: 'same day', valid_from: sameDay });
await runPhaseConsolidate(engine, {});
// All three valid_from values are equal; the (id ASC) tiebreaker
// makes the lowest-id row the "oldest" chronologically. Pin that
// contract since the trajectory CLI depends on this ordering.
const rows = await engine.executeRaw<{ id: number; valid_until: Date | null }>(
`SELECT id, valid_until FROM facts WHERE entity_slug = 'cdx4-acme-sameday' ORDER BY id ASC`,
);
expect(rows.length).toBe(3);
expect(rows[0].id).toBe(idA);
expect(rows[1].id).toBe(idB);
expect(rows[2].id).toBe(idC);
// First two are "older" by tiebreaker → both get valid_until set
// (= sameDay, since the next-newer fact has the same valid_from).
expect(rows[0].valid_until).not.toBeNull();
expect(rows[1].valid_until).not.toBeNull();
// Newest by tiebreaker stays NULL.
expect(rows[2].valid_until).toBeNull();
});
});
describe('R4b / R7 — cycle idempotency: re-run consolidate produces zero new takes (Codex F4 fix)', () => {
test('semantic upsert: second consolidate on identical state produces zero NEW takes', async () => {
await seedPage('cdx4-idempo-1');
const oldDate = new Date(Date.now() - 30 * 60 * 60 * 1000);
for (let i = 0; i < 4; i++) {
await insertFact({
entity_slug: 'cdx4-idempo-1',
text: 'stable claim',
valid_from: new Date(oldDate.getTime() + i * 60 * 60 * 1000),
});
}
// First run: 1 take, 4 facts consolidated.
const r1 = await runPhaseConsolidate(engine, {});
expect(r1.details.takes_written).toBe(1);
const countAfter1 = await engine.executeRaw<{ n: string }>(
`SELECT COUNT(*)::text AS n FROM takes WHERE page_id = (SELECT id FROM pages WHERE slug = 'cdx4-idempo-1')`,
);
expect(parseInt(countAfter1[0].n, 10)).toBe(1);
// Simulate the Codex F4 scenario: clear consolidated_at on every fact
// (extract_facts cycle phase wipes facts via delete-then-insert, which
// is functionally identical to NULL-ing consolidated_at). DO NOT touch
// valid_until — the prior consolidate wrote it; the semantic upsert
// should still find the take.
await engine.executeRaw(
`UPDATE facts SET consolidated_at = NULL, consolidated_into = NULL
WHERE entity_slug = 'cdx4-idempo-1'`,
);
// Second run: must NOT append another take.
const r2 = await runPhaseConsolidate(engine, {});
expect(r2.details.facts_consolidated).toBe(4);
// takes_written reports the NEW takes inserted this run; on the upsert
// hit path it's 0 (no new INSERT) but facts still get marked consolidated.
expect(r2.details.takes_written).toBe(0);
const countAfter2 = await engine.executeRaw<{ n: string }>(
`SELECT COUNT(*)::text AS n FROM takes WHERE page_id = (SELECT id FROM pages WHERE slug = 'cdx4-idempo-1')`,
);
expect(parseInt(countAfter2[0].n, 10)).toBe(1); // STILL 1 — no duplicate
// Facts were re-consolidated into the existing take.
const facts = await engine.executeRaw<{ consolidated_into: number }>(
`SELECT consolidated_into FROM facts WHERE entity_slug = 'cdx4-idempo-1' AND consolidated_into IS NOT NULL`,
);
expect(facts.length).toBe(4);
});
test('valid_until idempotency: second run leaves valid_until unchanged (no diff)', async () => {
await seedPage('cdx4-idempo-2');
const t1 = new Date('2026-01-15T00:00:00Z');
const t2 = new Date('2026-04-12T00:00:00Z');
const t3 = new Date('2026-07-08T00:00:00Z');
await insertFact({ entity_slug: 'cdx4-idempo-2', text: 'iterable', valid_from: t1 });
await insertFact({ entity_slug: 'cdx4-idempo-2', text: 'iterable', valid_from: t2 });
await insertFact({ entity_slug: 'cdx4-idempo-2', text: 'iterable', valid_from: t3 });
await runPhaseConsolidate(engine, {});
const before = await engine.executeRaw<{ id: number; valid_until: Date | null }>(
`SELECT id, valid_until FROM facts WHERE entity_slug = 'cdx4-idempo-2' ORDER BY valid_from ASC`,
);
// Reset consolidated_at to simulate extract_facts re-run.
await engine.executeRaw(
`UPDATE facts SET consolidated_at = NULL, consolidated_into = NULL
WHERE entity_slug = 'cdx4-idempo-2'`,
);
await runPhaseConsolidate(engine, {});
const after = await engine.executeRaw<{ id: number; valid_until: Date | null }>(
`SELECT id, valid_until FROM facts WHERE entity_slug = 'cdx4-idempo-2' ORDER BY valid_from ASC`,
);
// Same valid_until values; the IS DISTINCT FROM guard avoided rewrites.
expect(after.length).toBe(3);
for (let i = 0; i < before.length; i++) {
expect(after[i].id).toBe(before[i].id);
const a = after[i].valid_until ? new Date(after[i].valid_until!).toISOString() : null;
const b = before[i].valid_until ? new Date(before[i].valid_until!).toISOString() : null;
expect(a).toBe(b);
}
});
});