mirror of
https://github.com/garrytan/gstack.git
synced 2026-10-02 17:40:02 +02:00
garrytan
4973f96d98
The full startup workflow runs 1–3 real spec-review rounds (~280s each) and hit its 1200s capture in run 36385945043 at finalize. Review depth is the product's loop, so the case cannot fit a blocking lane without cutting rounds. It is now marathon tier with every assertion unchanged. skill-e2e-office-hours-design-draft.test.ts (periodic) runs the same fixed interview only through the Write that creates the design (269s in that run) and applies the full validator's design-draft checks, the required section reads and the launch/foreign-skill-read guards. validateOfficeHoursDesignDraft is extracted from validateOfficeHoursCompletion, which still applies it. Selection: office-hours-design-draft is registered periodic; the marathon-only file is already excluded from the gate and periodic plans by the B5 planner rule. Tier-alignment regexes and the valid-tier check accept 'marathon'. A type-only cast in plan-scope-selection.test.ts removes a diagnostic whose union print order made the ratchet identity unstable; baseline tightened.
69 lines
3.9 KiB
TypeScript
69 lines
3.9 KiB
TypeScript
/**
|
|
* Office-hours design-draft checkpoint (periodic). The fixed startup interview
|
|
* from CARVE_GUARDS['office-hours'] runs only through the Write that creates the
|
|
* design (observed at 269s of the full workflow in run 36385945043). It applies
|
|
* the full workflow's design-draft checks, required section reads and
|
|
* launch/skill-read guards. The complete review, approval and handoff workflow
|
|
* is marathon tier in skill-e2e-office-hours-section-loading.test.ts.
|
|
*/
|
|
import { test, expect } from 'bun:test';
|
|
import * as fs from 'node:fs';
|
|
import * as path from 'node:path';
|
|
import { describeE2ETier } from './helpers/e2e-gate';
|
|
import { CAPTURE_LONG_MS } from './helpers/eval-budgets';
|
|
import { setupSkillDir, skillFromWorktree, captureSectionReads, LONG_SECTION_CAPTURE_MS } from './helpers/auq-sdk-capture';
|
|
import { CARVE_GUARDS } from './helpers/carve-guards';
|
|
import { validateOfficeHoursDesignDraft } from './helpers/office-hours-completion';
|
|
|
|
const describePeriodic = describeE2ETier('periodic');
|
|
|
|
describePeriodic('/office-hours design-draft checkpoint (periodic)', () => {
|
|
test('the startup interview reads its sections and writes a complete design after an independent opinion', async () => {
|
|
const guard = CARVE_GUARDS['office-hours'];
|
|
const answersStart = guard.scenario.indexOf('This is a synthetic interview fixture');
|
|
const answersEnd = guard.scenario.indexOf('Use the available Agent tool');
|
|
if (answersStart < 0 || answersEnd < answersStart) {
|
|
throw new Error('fixture drift: CARVE_GUARDS office-hours scenario no longer has its interview block');
|
|
}
|
|
const { skillMd, sectionsFrom } = skillFromWorktree(guard.skill);
|
|
const planDir = setupSkillDir({
|
|
skillName: guard.skill, skillMd, sectionsFrom, tmpPrefix: 'gstack-office-hours-draft-',
|
|
});
|
|
const capture = await captureSectionReads({
|
|
planDir, skillName: guard.skill,
|
|
scenario: `Run office hours in STARTUP mode through the diagnostic, premise challenge,
|
|
alternatives, independent opinion and the design document.
|
|
${guard.scenario.slice(answersStart, answersEnd).trim()}
|
|
|
|
Use the available Agent tool for the independent opinion when the workflow calls
|
|
for it. It is read-only. An automated test is not a reason to skip it. This fixture
|
|
checks a separate real tool call: obtain the independent opinion on RosterCheck
|
|
before writing the design. A self-authored "second opinion" is not a subagent result.
|
|
Save the design to docs/designs/roster-check.md in this fixture; keep all artifacts
|
|
inside the fixture.
|
|
|
|
Checkpoint for this run: stop after the Write that saves the complete design
|
|
document. Do not run the spec review, approval, relationship closing or handoff;
|
|
the full workflow covers them separately. Then write REPORT.md with the design path
|
|
and a one-line outcome for each phase you completed.`,
|
|
testName: 'office-hours-design-draft', runId: `office-hours-design-draft-${process.env.EVALS_RUN_ID ?? 'local'}`,
|
|
timeout: LONG_SECTION_CAPTURE_MS,
|
|
});
|
|
const designPath = path.join(planDir, 'docs/designs/roster-check.md');
|
|
expect(capture.exitReason).toBe('success');
|
|
validateOfficeHoursDesignDraft({
|
|
designPath, toolCalls: capture.toolCalls,
|
|
designContent: fs.existsSync(designPath) ? fs.readFileSync(designPath, 'utf-8') : null,
|
|
});
|
|
const missing = guard.requiredReads.filter(section => !capture.readSections.has(section));
|
|
expect({ reportProduced: capture.reportProduced, read: [...capture.readSections], missing }).toEqual({
|
|
reportProduced: true, read: expect.any(Array), missing: [],
|
|
});
|
|
expect(capture.toolCalls.filter(call => call.tool === 'Skill')).toEqual([]);
|
|
const ownSkillPath = path.join(planDir, guard.skill, 'SKILL.md');
|
|
expect(capture.toolCalls.filter(call => call.tool === 'Read'
|
|
&& /(?:^|\/)SKILL\.md$/.test(String(call.input?.file_path ?? ''))
|
|
&& path.resolve(planDir, call.input.file_path) !== ownSkillPath)).toEqual([]);
|
|
}, CAPTURE_LONG_MS);
|
|
});
|