claude

solo-review

Solo code-review system: evals, hooks, a builder/verifier/reviewer loop, a first-time-user test and a bug-fix skill.

0 on GitHubv1.0.0updated 13 days ago
/

Plugin solo-review from marketplace solo-review-local in az9713/claude-code-review-solo. Commands read from the repo's manifest — 12 days ago.

This plugin runs code on your machine — its hooks fire automatically on events like tool use. The exact commands are listed below; read them before installing.

What it installs

Skills (6)

  • first-time-user

    Starts the app and sends a fresh agent that has never seen it to use it like a first-time customer, then lists every place that person got stuck. Use after a feature passes its evals, or when the user says "test it like a new user" or "where would a new user get stuck".

  • fix-a-bug

    Takes a bug report ("feature X does not work"), reproduces it, finds the real cause, and fixes it with a regression test. Use when the user says something is broken, not working, or regressed, or runs /solo-review:fix-a-bug.

  • review-loop

    Builds a reviewed feature and checks it in rounds with fresh-context subagents (builder, verifier, reviewer) until every eval passes. Use when the user runs /solo-review:review-loop FR-NN or says "build and review this feature". The feature must already have approved evals (status reviewed).

  • risk-level

    Rates a feature low, medium or high risk before it is built. Use before write-evals, or when the user asks "how risky is this feature", "how careful should I be", or "how strict should the review be".

  • setup-review

    One-time setup of the solo review system in a project. Creates REVIEW.md, links it from CLAUDE.md, makes the specs/ folder, and sets the git rules. Use when the user says "set up review", "install the review system", or starts using solo-review in a new project.

  • write-evals

    Turns a feature into an eval.md (checks an agent can run) and an HTML mockup, then asks the user to review them before anything is built. Use whenever the user brings a new feature to build, or says "write evals", "spec this feature", or "what should we check for this feature". This skill only write

Subagents (4)

  • builder

    Builds one feature from its RD.md, eval.md and mockup.html (round 1), then fixes items from a fix list (later rounds). Used only by the review-loop skill.

  • first-time-user

    Uses a live web app in the browser as a person who has never seen it, and reports every place where that person would get stuck. Cannot read code, docs or run commands. Used only by the first-time-user skill.

  • reviewer

    Reads a diff against the review checklist in REVIEW.md and reports problems the evals do not cover. Read-only. Used only by the review-loop skill.

  • verifier

    Runs every check in eval.md against a production build and writes PASS or FAIL with evidence into eval.md. Used only by the review-loop skill.

Hooks (1)

  • PreToolUse

    node

View source on GitHub

Maintain solo-review?

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 pluginslisted