feat(evals): with-skill vs without-skill arm benchmark — measures whether gstack's behavioral layer earns its tokens

Ponytail's honest-benchmark method pointed at gstack itself: 3 build-shaped
tasks (native-platform over-build trap, CRUD endpoint, bug fix with planted
decoys) x 2 arms, real claude -p sessions, scored on the git diff left
behind. A research instrument, not a release gate — no assertion compares
arm scores.

Arms use the PROVEN project-scope pattern: the with-arm installs a
build-discipline skill (extracted reuse-ladder + bounded-closer content, not
whole-file copies) into the fixture's .claude/skills/ with a CLAUDE.md
routing line and an explicit invocation; a live spike confirmed claude -p
discovers and invokes project-scope skills via the Skill tool (3 turns,
exact-output probe). Fixtures are git init + local bare origin; diff capture
is three lines of git, no worktree machinery.

Failure taxonomy: zero-diff arms are VALID scored cells (deterministic
0/none, no API call), harvest failures record harvest:null, judge_error
cells are excluded from aggregates but named in the report — nothing drops
silently. armJudge: fixed sonnet judge, 0-3 unrequested-structure rubric,
must name the construct or say none, bounded retry-on-malformed; callJudge
gains optional temperature/max_tokens (defaults unchanged). recordE2E now
populates tokens_used for every E2E. Eval schema v2: harvest gains
{insertions, deletions, net}, tolerant reads keep v1 runs comparable.

Registered periodic in E2E_TIERS + touchfiles (with the auq-repetition-cut
A/B); periodic detach timeout raised to the new shard-census floor. Free
selftest (8 tests, zero API) pins fixtures, extraction, arm asymmetry, diff
capture, judge plumbing, and the retry bound.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Garry Tan
2026-08-28 02:07:56 +00:00
co-authored by Claude Fable 5
parent 781f46d025
commit 4c20eca33b
21 changed files with 1076 additions and 10 deletions
+1 -1
View File
@@ -39,7 +39,7 @@
"eval:bg": "bin/gstack-detach --label evals --lock gstack-evals --timeout 5400 -- bun run test:evals",
"eval:bg:all": "bin/gstack-detach --label evals-all --lock gstack-evals --timeout 7200 -- bun run test:evals:all",
"eval:bg:gate": "bin/gstack-detach --label evals-gate --lock gstack-evals --timeout 25200 -- bun run test:gate:sharded",
"eval:bg:periodic": "bin/gstack-detach --label evals-periodic --lock gstack-evals --timeout 32400 -- bun run test:periodic:sharded",
"eval:bg:periodic": "bin/gstack-detach --label evals-periodic --lock gstack-evals --timeout 34200 -- bun run test:periodic:sharded",
"eval:list": "bun run scripts/eval-list.ts",
"eval:compare": "bun run scripts/eval-compare.ts",
"eval:summary": "bun run scripts/eval-summary.ts",