code-review
How to review code; a pull request, feature branch, local changes etc.
50
22%
Does it follow best practices?
Impact
100%
1.14xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./.claude/skills/code-review/SKILL.mdQuality
Discovery
22%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
The description is too vague and reads more like a topic heading than a functional skill description. It lacks concrete actions (what specific review activities are performed), has no explicit 'Use when...' trigger clause, and misses common user-facing terms like 'PR' or 'diff'. The description needs significant improvement to help Claude reliably select this skill from a large pool.
Suggestions
Add specific concrete actions such as 'Analyzes code diffs, identifies bugs, checks for style issues, suggests improvements, and summarizes changes'.
Add an explicit 'Use when...' clause, e.g., 'Use when the user asks to review a PR, check a diff, review code changes, or analyze a feature branch'.
Include common trigger term variations like 'PR', 'merge request', 'diff', 'code review', 'review my changes', and 'staged changes'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description says 'how to review code' which is vague. It lists contexts (pull request, feature branch, local changes) but does not describe concrete actions like 'check for bugs, suggest improvements, analyze diffs, flag security issues'. | 1 / 3 |
Completeness | The description partially addresses 'what' (review code) but is very weak on specifics, and there is no explicit 'when' clause or trigger guidance. Per the rubric, a missing 'Use when...' clause caps completeness at 2, and the weak 'what' brings it down to 1. | 1 / 3 |
Trigger Term Quality | Includes some natural keywords users might say like 'pull request', 'feature branch', 'local changes', and 'review code', but misses common variations like 'PR', 'code review', 'diff', 'merge request', or 'review my changes'. | 2 / 3 |
Distinctiveness Conflict Risk | The mention of 'pull request, feature branch, local changes' provides some specificity to the code review domain, but 'review code' is broad enough to overlap with linting skills, static analysis skills, or general coding assistance skills. | 2 / 3 |
Total | 6 / 12 Passed |
Implementation
22%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is too vague and abstract to meaningfully guide Claude's behavior during code reviews. It lists generic review categories that Claude already knows without providing concrete examples, specific commands (e.g., gh pr diff), structured workflows, or example review outputs. The references to CLAUDE.md and the glossary skill are positive but insufficient to compensate for the lack of actionable content.
Suggestions
Add a concrete workflow: e.g., 1) Fetch the diff/PR, 2) Read changed files, 3) Check for specific issues, 4) Format review comments with examples of good vs bad feedback.
Include specific commands for common review scenarios (e.g., `gh pr diff <number>`, `git diff main...HEAD`) and example review comment output format.
Provide at least one concrete example of a review comment showing the expected level of detail and tone, rather than just saying 'be constructive and helpful'.
Remove the generic bullet list of review areas (code quality, bugs, performance, security, tests) since Claude already knows these, or replace it with project-specific criteria that add real value.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Reasonably brief but includes some generic advice ('Be constructive and helpful') that Claude already knows. The bullet list of review areas (code quality, bugs, performance, security, tests) is standard knowledge for any code reviewer and doesn't add much value. | 2 / 3 |
Actionability | The skill provides only vague, abstract direction ('provide feedback on code quality and best practices') with no concrete examples of what good feedback looks like, no specific commands for reviewing PRs or diffs, and no example input/output of a review comment. | 1 / 3 |
Workflow Clarity | There is no sequenced workflow—just a flat bullet list of topics to cover. No steps for how to approach a review (e.g., read diff first, check tests, then comment), no validation checkpoints, and no guidance on how to structure or deliver the review output. | 1 / 3 |
Progressive Disclosure | The skill references CLAUDE.md for style guidance and mentions invoking the glossary skill, which shows some awareness of external resources. However, there are no structured sections, no links to detailed review guidelines, and the content is a single flat block rather than well-organized sections. | 2 / 3 |
Total | 6 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- MetaMask/ocap-kernel
- Commit
a3eff0e
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.