[renmengye]

Moves the pinned review model (the [codex] model in harnesses.toml, #602) from gpt-5.6-terra to gpt-6.1-terra. Every caller that leaves model unset follows it: lab repos on @main at merge, @v0 callers at the next release.

Checked: codex exec -m gpt-6.1-terra answers with the lab's Codex login. A manual advisory-review-agent-manual run on this branch exercises the pinned Codex CLI (0.160.0) with the new model in CI.

Gate: ruff check, ruff format --check, mypy, full pytest pass.

Built by Claude Opus 5.5; reviewed by the advisory reviewer on this PR.

[github-actions]

Round 1 — reviewed head 35a4fabd — reviewer codex/gpt-6.1-terra.

Advisory findings from outerloop — the code owner decides. Reply to disagree; the outerloop:no-review label opts this PR out.

Verdict: no defects found.

No defects found. The manual CI run and model access with reviewer API keys could not be verified from the checkout.

[github-actions]

Round 1 — reviewed head 35a4fabd — reviewer summarizer:codex/gpt-5.6-terra over coverage+credentials+deployment+general+lifecycle+prose.

terra
Advisory findings from outerloop — the code owner decides. Reply to disagree; the outerloop:no-review label opts this PR out.

Verdict: nothing blocking — 1 advisory note.

Advisory (non-blocking):

• Suggestion: No test pins the new Codex review model. [coverage] The test compares the harness model to cli.pins("codex")["model"], so it passes if harnesses.toml still contains gpt-5.6-terra and does not protect the claimed gpt-6.1-terra change. (tests/test_review_agent_cli.py:77; high confidence)

One non-blocking coverage suggestion recorded. Rejected findings: none.