[polewskm]

The smoke eval now fails when evals/baseline-smoke.json is missing after it was committed, naming the deleting commit (it still passes with a notice before the first baseline; a shallow checkout fails closed, so the workflow fetches full history). New tests pin the smoke eval workflow's order and that evals/run.ts writes its output before failing a run over --max-cost, via an exported writeRunOutput.
The loop limit, open-issue synthesis and Autopilot: replies now use the engine's writerCheck (association, then a cached permission lookup that fails closed), so a private organization member counts; open-issue author lookups are capped at 20. The /review gate stays association-only, and the README tells private members how to use it.
Two merge steps bring in origin/main, so this pull request must merge with a merge commit, not by rebasing.

Closes #93
Closes #89

A first review round moved the shared writer-check test stub into one module.

[ncode-review]

🤖 NCode Review at d88b4de: 2 new findings on the changed lines.

Not on a changed line:

• 🟡 F2 Nit: No test covers when main() reads the git history evals/smoke.ts:165. main() looks up deletedIn only when the baseline path is the committed default and the file is missing, and no test covers that condition. Its parts (smokeCheck, checkFiles with deletedIn, lastBaselineCommit) are each tested. Fix: Optional: move the condition into a small exported function and test it with the default path, a --baseline path and a file that exists.

[ncode-review]

_{Superseded by the review at fd9d544.}

[polewskm]

/review

[ncode-review]

🤖 Review complete

Verdict: ✅ Approve — Re-review of what changed since the last review at d88b4de: 6 hunks in 4 files and changes taken back in 3 files, by correctness only. Earlier findings: F1 still open, F3 fixed. No further findings above nits.

1 reviewer run — 0 findings (run details)

Reviewer	Model	Findings	Top
correctness	sonnet-5-5	0	🟢

_{Trial: findings are posted as review threads and in the check run. $0.03 on Amazon Bedrock, 9s, at fd9d544. (how to interact)}

📋 Outcomes

Merged at fd9d544: 1 fixed, 1 dismissed, 0 acknowledged, 0 ignored.

Finding	Reviewer	Severity	Outcome
F1	architecture, tests	🟡 Nit	👎 dismissed
F3	hygiene	🟡 Nit	✅ fixed