CalvinBackup/gstack
1
0
Fork 0
mirror of https://github.com/garrytan/gstack.git synced 2026-09-17 02:15:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

test: registries, budgets and suite reconciled for Aside-first with the $B fallback

Browse Source 
Touchfiles + E2E tiers gain the Aside keys, coverage matrix and eval baselines updated, size budget re-baselined to parity-baseline-v1.80.0.0.json (the contract plus fallback ride in every browsing skill), parity ceilings ratcheted with measured values, LLM-judge prompts and the E2E fixtures speak Aside-first, browse-fallback.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
Sina
2026-09-05 16:48:38 -04:00
co-authored by Claude Fable 5.1
parent 993692cd54
commit 886217894f
21 changed files with 1281 additions and 166 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/skill-size-budget.test.ts
+15 -4
View File
@@ -14,7 +14,13 @@
* compares LIVE eval runs (tool calls, turns, cost); this one compares
* static SKILL.md sizes. Both gate-tier.
*
* The baseline lives at test/fixtures/parity-baseline-v1.69.1.0.json,
* Baseline rebased v1.69.1.0 → v1.80.0.0: the Aside-first browser contract
* ({{ASIDE_SETUP}}) plus the gstack-browser fallback block now ride in every
* browsing skill (~9KB), which pushed benchmark and scrape past 1.5× of the
* v1.69.1.0 anchor. Deliberate, corpus-wide, receipted in the v1.80.0.0
* CHANGELOG; the v1.69.1.0 fixture stays on disk for history.
*
* The previous baseline lived at test/fixtures/parity-baseline-v1.69.1.0.json,
* re-captured 2026-08-25 during token-reduction Phase 1 (bash consolidation
* moved ~11-13KB of inline preamble bash per skill into bin/gstack-skill-start
* and bin/gstack-skill-end — a deliberate corpus-wide shrink; receipt:
@@ -42,7 +48,7 @@ import { logBudgetOverride } from './helpers/budget-override';
import { CARVED_SKILLS } from './helpers/carve-guards';
const REPO_ROOT = path.resolve(import.meta.dir, '..');
const BASELINE_PATH = path.join(REPO_ROOT, 'test', 'fixtures', 'parity-baseline-v1.69.1.0.json');
const BASELINE_PATH = path.join(REPO_ROOT, 'test', 'fixtures', 'parity-baseline-v1.80.0.0.json');
// Default per-skill ratio is 1.50 (50% growth tolerance). Adjusted v1.52.0.0
// (cathedral cap audit) from 1.05 → 1.50: a 5% ratio tripped on legitimate
@@ -62,11 +68,11 @@ interface Regression {
}
describe('SKILL.md size budget regression (gate, free)', () => {
test('parity-baseline-v1.69.1.0.json exists', () => {
test('parity-baseline-v1.80.0.0.json exists', () => {
expect(fs.existsSync(BASELINE_PATH)).toBe(true);
});
test('no skill exceeds v1.69.1.0 baseline size × ratio', () => {
test('no skill exceeds v1.80.0.0 baseline size × ratio', () => {
const baseline: ParityBaseline = JSON.parse(fs.readFileSync(BASELINE_PATH, 'utf-8'));
const current = captureBaseline({ repoRoot: REPO_ROOT });
@@ -182,10 +188,15 @@ describe('SKILL.md size budget regression (gate, free)', () => {
// preamble-tier frontmatter fell through `?? 4`). Their tiers are now
// declared correctly (1-2), shedding the tier-2..4 onboarding prose they
// never should have carried (-271 lines each for tier 1).
// - browse: the baseline measured the headless-browse skill with its ~17 KB
// $B command reference + snapshot-flag tables inline. /browse drives the
// Aside browser first and carries the Aside contract in the skeleton; the
// command tables live in the carved browse/sections/command-list.md.
const INTENTIONAL_SHRINKS = new Set<string>([
'spec',
'scrape', 'diagram', 'open-gstack-browser',
'landing-report', 'pair-agent', 'skillify',
'browse',
]);
const undershoots: Array<{