Files
gstack/test/autoplan-pairing.test.ts
Garry Tanandcapy-ai[bot] 90a170c941 v1.92.50.0 wave(autoplan): #3781 #3743 #3751 #3762 (#3788)
* fix(eval-plan): per-arm pilot projection and --seat-metered arms (#3653)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): guard tolerates Claude Code's cwd-reset footer and recovers after an unreadable init (#3651)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (journal reader strips the cwd-reset footer; an init with no text block no longer locks the session)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: journal reader skips a failed or unreadable init, as invocation() does

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* feat(autoplan): --ceo-only runs the CEO review alone with zero questions (GST-537)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(eval-plan): all-seat pilot, single-attempt arm and --seat-metered CLI wiring

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(eval-plan): summed deviation adds arm deviations, matching the summed band (arms share rows)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (test pins an unpriced seat-metered arm)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (seat-metered declaration fixed per --out, price row still required, a real charge settles as spent)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: classify autoplan-ceo-only in the Codex host metadata

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: a seat-metered arm that settles a charge is projected as a priced arm

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: select autoplan-ceo-only on the skills it runs; say it needs Claude Code

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(skills): /eval-plan --seat-metered and per-arm projection

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(eval-plan): skill prose stays under its context budget and carries no $N token

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: run autoplan-ceo-only in the marathon lane

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(skills): a seat-metered arm leaves the seat set once it settles a charge

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: changelog entry for /autoplan --ceo-only

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(autoplan): init_failed covers unreadable init output and the cwd-reset line (#3651)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: count the autoplan-ceo-only paid file in the shard inventory

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(autoplan): #3651 tests pass the test-code type ratchet

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(dx): CAP finding titles carry a one-line description (GBRA-223 claim 2)

The DX error contract told both autoplan DX voices to title capped findings
exactly CAP4:<path>:<item>, so gates, tasks and reconcile saw a bare code.
The title is now CAP4:<path>:<item> <what fails>; the grader already parses
that shape.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): reconcile pairs same-file findings that carry no line (GBRA-223 claim 1)

Plan reviews name files, rarely lines, and reconcile associated by location
only when both findings had a line, so findings both voices raised on one
file were filed new and native-only. A same-file pair with a missing line is
now associated as unverified (basis file), never confirmed; a CAP code no
longer counts in title comparison. DRIVER.md names submit --reconcile.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): open Critical and High findings reach the gate as apply items (GBRA-223 claim 3)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): a spend cap binds on every unattended host; --estimate-usd persists; DRIVER names --usage-usd (GBRA-223 claim 4)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): run binds complete api results, names a truncated response, --max-tokens; DRIVER documents run (GBRA-223 claim 6)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* feat(autoplan): carry an item-by-item answer when the later gate asks nothing new (GBRA-223 claim 7)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(dx): trim the compact error contract so plan-devex-review stays under its size pin

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (p1b applies finding fixes, api attempts reserve no host default, cross-file title pairing, paired severity counted)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: review fixes (priced api reservations, no file-only twins, two-pass pairing, partial replies keep p1-dependent defaults open, outside refusal degrades)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* P7: record whose words answered a gate (--words-by agent:<id>), never carry an agent's own words, import each row under its own answerer (GEVA-2)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: review pass 2 (p1b needs --rec-hash after recommend; api calls send the output limit they reserve for)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* changelog: fragment for the /autoplan field-report wave (GST-672)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: captain record notes --words-by agent attribution

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: follow finding apply items in approval-class and Captain flow tests; bind run-api limit to DEFAULT_MAX_TOKENS

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* wave autoplan: lower the gemini plan-devex-review pin to its exact size after #3781's DX trim

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): CAP titles on different paths never match by title in reconcile or confirm rounds (wave review)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): a truncated api call settles its cost and a retry is admitted afresh against the cap

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): an applied finding stays open until the revised plan is reviewed; a shared location alone never twins two findings

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(approval): a plan gate answered in an agent's own words is not owner approval

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(gate): a new answer replaces the earlier answer's authorship and words

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): a failed api call with no usage keeps its attempt open; reservations price the bound prompt (wave fix-check)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): a spend cap fails closed: only a call refused before sending keeps its attempt; a sent call with unknown usage settles at its reservation

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(autoplan): the 16384 output default applies to both providers; changelog for the wave's integrator fixes

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(autoplan): how run settles a failed api call; changelog scopes it to the native voice

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(autoplan): a spend cap fails closed on every exit of run; an unpriced model reserves the default once a cap is set

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(autoplan): every sent call settles; an unpriced model under a cap reserves the default

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(autoplan): DRIVER and help say a cap reserves in any session kind and how run settles a sent call

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

---------

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
2026-11-02 01:35:47 +00:00

121 lines
9.1 KiB
TypeScript

/**
* Pairing negative control (next-wave B2/B-a): finding pairs copied verbatim
* from a real autoplan run. File:line proximity associates findings but never
* confirms them, and neither does a title match alone: a title match with
* the same remedy confirms, opposite remedies disagree, different remedies stay
* unverified; the parent's override with one shared proposal also confirms.
*/
import { describe, expect, test } from 'bun:test';
import fixture from './fixtures/autoplan-pairing/run-20261014-pairs.json';
import { reconcile, titleScore } from '../lib/autoplan-reconcile';
import { challengeDefault } from '../lib/autoplan-gate';
const pair = (name: string) => fixture.pairs.find(p => p.name.startsWith(name))! as any;
const challenge = (rows: any[], changes = 'direction') => rows.map(r => ({ ...r, user_challenge: true, changes }));
describe('pairing negative control from a real run', () => {
test('two findings three lines apart on one file are associated but never confirmed, and never recommend b', () => {
const p = pair('ceo r1 outside-1');
const [row] = reconcile(p.native, p.outside).rows;
expect(row).toMatchObject({ disposition: 'unverified', basis: 'file_line', native_id: p.native[0].id });
expect(row!.resolution).toStartWith('same location, agreement unverified (bin/gstack-capy-install:197)');
const [n] = challenge(p.native), [o] = challenge(p.outside);
expect(challengeDefault(row!, n, o)).toMatchObject({ recommended: 'a', basis: 'plan direction; same location, agreement unverified' });
});
test('a location match with different titles stays unverified until the parent confirms it with one proposal', () => {
const p = pair('dx r2 outside-2');
const [row] = reconcile(p.native, p.outside).rows;
expect(row!.disposition).toBe('unverified');
const [confirmed] = reconcile(p.native, p.outside, [{ outside_id: p.outside[0].id, native_id: p.native[0].id, disposition: 'confirmed', resolution: 'same trailer grammar break', proposal: 'keep `gstack: fail <components>` and specify the pins exit' }]).rows;
expect(challengeDefault(confirmed!, challenge(p.native, 'scope')[0], challenge(p.outside, 'scope')[0])).toMatchObject({ recommended: 'b', basis: 'both voices; shared proposal' });
});
test('the same problem on different files is new until the parent associates it; different remedies keep a', () => {
const p = pair('dx r2 native-2');
expect(reconcile(p.native, p.outside).rows[0]).toMatchObject({ disposition: 'new', native_id: null });
const [row] = reconcile(p.native, p.outside, [{ outside_id: p.outside[0].id, native_id: p.native[0].id, disposition: 'confirmed', resolution: 'both: a recommendation edit changes what a seen reply resolves to' }]).rows;
expect(challengeDefault(row!, challenge(p.native, 'scope')[0], challenge(p.outside, 'scope')[0])).toMatchObject({ recommended: 'a', basis: 'plan direction; remedies differ; no shared proposal' });
});
test('the same title with opposite remedies disagrees, as an ordinary finding and as a user challenge; never b', () => {
const p = pair('constructed same title, opposite');
const [row] = reconcile(p.native, p.outside).rows;
expect(row).toMatchObject({ disposition: 'disagree', basis: 'title' });
expect(row!.resolution).toStartWith('remedies conflict (same title): native "move attempts into a new retry_attempts table"');
const [n] = challenge(p.native, 'scope'), [o] = challenge(p.outside, 'scope');
const [crow] = reconcile([n], [o]).rows;
expect(crow!.disposition).toBe('disagree');
expect(challengeDefault(crow!, n, o)).toMatchObject({ recommended: 'a', basis: 'plan direction; voices disagree' });
});
test('the same title with different remedies is only associated; the same remedy confirms and recommends b', () => {
const differ = pair('constructed same title, different');
const [row] = reconcile(differ.native, differ.outside).rows;
expect(row).toMatchObject({ disposition: 'unverified' });
expect(row!.resolution).toStartWith('same title, remedies differ');
expect(challengeDefault(row!, challenge(differ.native, 'scope')[0], challenge(differ.outside, 'scope')[0])).toMatchObject({ recommended: 'a', basis: 'plan direction; remedies differ; no shared proposal' });
const same = pair('constructed same title, same remedy');
const [ok] = reconcile(same.native, same.outside).rows;
expect(ok).toMatchObject({ disposition: 'confirmed', basis: 'title' });
expect(challengeDefault(ok!, challenge(same.native, 'scope')[0], challenge(same.outside, 'scope')[0])).toMatchObject({ recommended: 'b', basis: 'both voices; same remedy' });
});
});
describe('plan reviews name a file without a line (GBRA-223 claim 1)', () => {
const row = (voice: 'native' | 'outside', n: number, f: Record<string, unknown>) => ({ schema_version: 1, run: 'r', id: `r-dx-${voice}-${n}`, phase: 'dx', voice, disposition: 'open', ...f }) as any;
const native = row('native', 1, { severity: 'Critical', title: 'per-turn resolver call has no hard timeout or fail-open', file: 'src/resolver.ts', resolution: 'add a 2s timeout and fail open' });
test('a CAP-coded outside title on the same file is associated, unverified, never new', () => {
const outside = row('outside', 1, { severity: 'Critical', title: 'CAP4:src/resolver.ts:recovery', file: 'src/resolver.ts', resolution: 'bound the call' });
const { rows, native_only } = reconcile([native], [outside]);
expect(rows[0]).toMatchObject({ disposition: 'unverified', basis: 'file', native_id: native.id });
expect(rows[0]!.resolution).toStartWith('same file, agreement unverified (src/resolver.ts)');
expect(native_only).toEqual([]);
});
test('differently worded titles on the same file associate; the best title among same-file natives wins', () => {
const other = row('native', 2, { severity: 'Medium', title: 'resolver output is not cached between turns', file: 'src/resolver.ts' });
const outside = row('outside', 1, { severity: 'Critical', title: 'Resolver call has no timeout and can hang the turn', file: 'src/resolver.ts' });
expect(reconcile([other, native], [outside]).rows[0]).toMatchObject({ disposition: 'unverified', basis: 'file', native_id: native.id });
});
test('a CAP code is not part of the title comparison', () => {
const outside = row('outside', 1, { severity: 'Critical', title: 'CAP4:src/resolver.ts:recovery per-turn resolver call has no hard timeout or fail-open', file: 'src/resolver.ts', resolution: 'add a 2s timeout and fail open' });
expect(reconcile([native], [outside]).rows[0]).toMatchObject({ disposition: 'confirmed', basis: 'title' });
});
test('controls: two lines far apart, or different files, stay new', () => {
const lined = row('native', 1, { ...native, id: 'r-dx-native-1', line: 10 });
expect(reconcile([lined], [row('outside', 1, { severity: 'High', title: 'retry loop has no bound', file: 'src/resolver.ts', line: 90 })]).rows[0]!.disposition).toBe('new');
expect(reconcile([native], [row('outside', 1, { severity: 'High', title: 'CAP6:src/cli.ts:fix', file: 'src/cli.ts' })]).rows[0]!.disposition).toBe('new');
});
test('a title twin is placed before the same-file fallback, so an unrelated finding cannot take its native', () => {
const n1 = row('native', 1, { severity: 'High', title: 'token refresh race drops the session', file: 'plan.md' });
const o1 = row('outside', 1, { severity: 'High', title: 'migration lacks rollback', file: 'plan.md' });
const o2 = row('outside', 2, { severity: 'High', title: 'token refresh race drops the session', file: 'plan.md' });
const { rows } = reconcile([n1], [o1, o2]);
expect(rows.map(r => [r.outside_id, r.native_id, r.basis])).toEqual([[o1.id, null, 'none'], [o2.id, n1.id, 'title']]);
});
test('control: the same CAP item and wording on two different files never pairs by title', () => {
const a = row('native', 1, { severity: 'High', title: 'CAP6:bin/gstack-upgrade:recovery no retry bound on network failure', file: 'bin/gstack-upgrade' });
const b = row('outside', 1, { severity: 'High', title: 'CAP6:lib/telemetry-sync.ts:recovery no retry bound on network failure', file: 'lib/telemetry-sync.ts' });
expect(reconcile([a], [b]).rows[0]).toMatchObject({ disposition: 'new', native_id: null });
});
test('two CAP titles on different paths never match by title, even when both findings sit on the plan file', () => {
const A = 'CAP6:bin/gstack-eval-plan pilot start:fix no fix line names the command';
const B = 'CAP6:bin/gstack-outside-voice run:fix no fix line names the command';
expect(titleScore(A, B)).toBe(0);
expect(titleScore(A, A)).toBe(1);
expect(titleScore(A, 'no fix line names the command')).toBe(1);
const n = row('native', 1, { severity: 'High', title: A, file: 'plan.md', resolution: 'name the command' });
const o = row('outside', 1, { severity: 'High', title: B, file: 'plan.md', resolution: 'name the command' });
expect(reconcile([n], [o]).rows[0]).toMatchObject({ basis: 'file', disposition: 'unverified' });
const bare = (x: any) => { const { file, ...rest } = x; return rest; };
expect(reconcile([bare(n)], [bare(o)]).rows[0]).toMatchObject({ disposition: 'new', native_id: null });
});
});