Agent
Read-only branch-level documentation drift review. Runs once before final review when documented behavior or authority changes, or explicitly as a whole-repo audit.
Agent
Read-only branch-level documentation drift review. Runs once before final review when documented behavior or authority changes, or explicitly as a whole-repo audit.
Agent
Runs a complete orchestrator-supplied inspection task on an explicitly selected model. Never mutates the repository.
Agent
One deep review of a whole branch before it merges, on the strongest model you have. Judges the branch as a single change against what was asked, not as the commits it arrived in. Never edits code. Runs once per branch, after the per-commit reviewers have done their passes.
Agent
Implements an approved plan, one slice at a time. Dumb code over clever code. Stops and reports when the plan contradicts the real code, instead of improvising a different design.
Agent
Adversarial spec review before a commit: does the change do what was asked — all of it, and nothing more? Never edits code. Runs together with the taste-reviewer before every commit, except trivial ones.
Agent
Taste review of the working diff before a commit: does the code follow the repo's standards? Never edits code. Runs together with the spec-reviewer before every commit, except trivial ones.
Agent
Adversarial review of whether a diff has the necessary, effective, independent, and proportionate automated evidence. Never edits code. Runs conditionally for behavior, test, test-infrastructure, and bug-fix diffs.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: