mcpbeat

Claude Review

redis/claude-review

Use when asked to independently review Claude-authored docs or code changes, Claude-generated diffs, or Claude PR work in the Redis docs repo.

3k tokens
context cost
the whole folder, loaded on every use
3
files
instructions only
0
copies elsewhere
how many repositories repackaged it
76
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/redis/docs --skill claude-review

What comes with it

9 808 bytes besides the instruction
references/claude-review-patterns.md
references/review-style.md

The instruction itself

5 sections, as written by the author

Claude Review

Use this skill for independent reviews of Claude-authored work in this repo. The goal is to find real correctness, docs, consistency, test, and maintainability issues without becoming biased by past Claude behavior.

Required References

Read these before reviewing:

  • references/review-style.md for output format and severity.
  • references/claude-review-patterns.md for recurring review patterns.

Treat the pattern file as a map of places to inspect, not as evidence. The current diff and current workspace always win.

Workflow

  • Identify the review target:
  • Prefer the user-supplied diff, PR, branch, or file list.
  • If the target is local, inspect git diff, git status --short, and the touched files before judging.
  • If the target is a PR, inspect the PR diff and any relevant review comments.
  • Read the nearby source of truth:
  • Adjacent docs pages, examples, tests, config, data files, shortcodes, generated mappings, or skill instructions touched by the change.
  • Existing repo conventions before claiming a convention was violated.
  • Use claude-review-patterns.md as a targeted checklist:
  • Pick only patterns relevant to the touched files and change type.
  • For each suspected issue, verify the exact current file and line before reporting it.
  • Do not report a pattern match unless it causes an actionable problem in this change.
  • Produce findings first:
  • Order by severity.
  • Include precise file/line references.
  • Explain the user-visible or maintainer-visible impact.
  • Keep summaries secondary and brief.
  • If no issues are found:
  • Say that clearly.
  • Mention residual risk, unrun tests, or review limits.

Independence Rules

  • Historical patterns suggest what to inspect; they never prove a current bug.
  • Never assume Claude made a mistake because Claude made a similar mistake before.
  • Verify against the current workspace to avoid stale-snapshot findings.
  • Prefer fewer, stronger findings over broad speculative concerns.
  • Do not modify this skill or its references during a review unless the user explicitly asks.

Skill Memory Updates

At the end of a review, add a short Suggested skill-memory updates section only if the review revealed a reusable pattern that is:

  • concrete enough to scan for in future reviews;
  • backed by a current finding or a clearly described false positive;
  • not already covered by references/claude-review-patterns.md.

Suggest the update text, but do not apply it unless explicitly asked. New entries should follow the schema in claude-review-patterns.md.

How to use it

Copy the folder

Take redis/claude-review from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.