[polewskm]

This PR acts on the 2026-11-01 evals (9 runs through the Eval and Smoke eval workflows, $24.83 in total) and prepares v1.0.0:

• Add a verification stage that proves or refutes each blocker and should-fix #41: verification is now on by default (CODE_REVIEW_VERIFICATION=off turns it off). On the smoke subset it caught the same 2/3, with 0 must-not flags, at 5 and 5 blockers and should-fixes against 7 and 8 without it, for about 3% more cost.
• Measure synthesis at a lower effort or on Sonnet #135: synthesis stays on Opus at the default effort. Sonnet halved synthesis's cost but posted 12 blockers and should-fixes against Opus's 7–8, and medium effort saved nothing.
• Run the re-review evals and record what they show #123: the re-review cases cost $0.45 and $0.12 in total, posted no findings, and merged 0 and counted 0 near findings, so mergeOverlapping stays as it is. All of these results are recorded in the README's Evals section.
• Run a cost-capped smoke eval when a prompt, charter or model changes #52: commits the first evals/baseline-smoke.json, from the smoke run on main after Dispatchable evals, synthesis knobs and the decisions rule #140's decisions-rule fix (caught 2/3, must-not quiet).
• Roll out the 2026-10-31 changes: release, consumer pins, app permission, CodeQL #97's hand count: the October report's totals match the markers exactly (12 PRs, 20 fixed, 34 dismissed), but Test eval review cost and explain's answer #115 showed the outcome rules misreading the autopilot's own replies. A writer's latest Autopilot: fixed in <sha> (when that commit is in the PR after the review) or Autopilot: not changed now decides.
• Release v1.0.0 and freeze the caller–engine interface #57: the Unreleased entries move into ## [1.0.0] - 2026-11-01, and the README states the v1 contract and how Dependabot proposes v1 bumps. The Release workflow cuts v1.0.0 after this merges.

Closes #41
Closes #123
Closes #135
Refs #52
Refs #97
Refs #57

[ncode-review]

🤖 NCode Review at 95d23d5: 5 new findings on the changed lines.

[ncode-review]

_{Superseded by the review at 035aa01.}

[polewskm]

/review

[ncode-review]

🤖 Review complete

Verdict: ✅ Approve — Re-review of what changed since the last review at 95d23d5: 6 hunks in 6 files and changes taken back in 3 files, by correctness only. Earlier findings: F1 fixed, F2 fixed, F3 fixed, F4 fixed, F5 still open. 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. $0.04 on Amazon Bedrock, 10s, at 035aa01. (how to interact)}

📋 Outcomes

Merged at 035aa01: 2 fixed, 3 dismissed, 0 acknowledged, 0 ignored.

Finding	Reviewer	Severity	Outcome
F1	rule-check	🟡 Nit	✅ fixed
F2	rule-check	🟡 Nit	✅ fixed
F3	tests	🟡 Nit	👎 dismissed
F4	tests	🟡 Nit	👎 dismissed
F5	hygiene	🟡 Nit	👎 dismissed