claude

code-review-max

Max-effort, recall-mode code review as a Kiro workflow for native Windows with no Python: eight parallel finder angles, per-candidate verification, a gap sweep, re-votes on unstable verdicts, then ranked findings and postable Conventional Comments.

1 on GitHubupdated 7 days ago
/

Plugin code-review-max from marketplace kiro-starter-kit in dwalleck/kiro-starter-kit. Commands read from the repo's manifest — 6 days ago.

What it installs

Subagents (4)

  • cr-clerk

    Bookkeeping agent for the code-review-max-windows workflow. Runs crtool, the review's deterministic data step, and makes the small judgment calls it cannot (which candidates are duplicates, how findings rank).

  • cr-commenter

    Writes the postable review comment for each reported finding of the code-review-max-windows workflow, in the Conventional Comments format. Chooses the label and decorations and writes the prose; never re-judges the finding.

  • cr-finder

    Code-review finder for the code-review-max-windows workflow. Reviews one diff from ONE assigned angle and writes candidate findings as JSON. Read-only except for its own candidates file.

  • cr-verifier

    Code-review verifier for the code-review-max-windows workflow. Takes ONE candidate finding off a queue, verifies it against the source, writes a three-state verdict, and updates the queue.

View source on GitHub

More from dwalleck/kiro-starter-kit

All 6 plugins →

Polyglot unit-test generation via an orchestrated Research → Plan → Implement → Validate agent pipeline, with deep .NET test-framework skills (MSTest/TUnit authoring, migrations, CRAP scores).

28 skills · 4 agents

1

Polyglot unit-test generation as a Kiro workflow (kiro-cli 2.26.0+, KAS engine) built on the current dotnet-test plugin: test-engineer entry agent, then Research → Plan → repeat[Implement, Validate] gated on a verdict file, handing off through .git/testagent/ so nothing is ever stageable. Ships the vendored dotnet-test skills.

11 skills · 6 agents

1

Builds features via an engine-enforced Spec → Probe → Design → Plan → Build → Validate pipeline that refuses to advance on a vague spec or disproven understanding, and loops when built code drifts from the plan.

1

Orchestrated PR code review agents for the Kiro CLI: an orchestrator plus six specialists (code-reviewer, code-simplifier, comment-analyzer, pr-test-analyzer, silent-failure-hunter, type-design-analyzer).

1

Orchestrated PR code review agents for the Kiro CLI: an orchestrator plus six specialists (code-reviewer, code-simplifier, comment-analyzer, pr-test-analyzer, silent-failure-hunter, type-design-analyzer).

1

Maintain code-review-max?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins6 plugins