AgentSkillsCN

Review Prompt

审查 Readme

SKILL.md

Review PROMPT.md

Review the Blog in a Prompt PROMPT.md file for correctness, consistency, and executability.

Trigger phrases

  • "review the prompt"
  • "check the prompt"
  • "audit PROMPT.md"
  • "validate the prompt"

What this skill does

Performs a comprehensive review of PROMPT.md to ensure it will execute correctly when read by Claude Code.

1. Structural consistency

  • Verify all section cross-references are correct (e.g., "see Section 7.0" actually points to the right content)
  • Check that numbered steps are sequential and none are missing
  • Verify all internal links between sections are valid

2. Technical accuracy

  • Verify CLI commands are correct and use valid flags
  • Check that version numbers are consistent (e.g., Next.js version in create-next-app matches package.json pinning)
  • Verify file paths mentioned match the required project structure
  • Check that code examples are syntactically correct

3. Logical flow

  • Verify the order of operations makes sense (e.g., can't reference files before they're created)
  • Check for circular dependencies or impossible sequences
  • Ensure prerequisites are mentioned before they're needed

4. Contradiction detection

  • Look for conflicting instructions (e.g., "do X" in one place, "don't do X" in another)
  • Check that timing instructions are consistent (e.g., when to create skills)
  • Verify allowlists/blocklists don't contradict each other

5. Completeness

  • Verify all mentioned files have creation instructions
  • Check that all user questions have handling instructions for responses
  • Ensure error scenarios have recovery paths

6. Executability

  • Identify any ambiguous instructions that could be interpreted multiple ways
  • Flag instructions that assume context not provided
  • Check for missing details that would block execution

Output format

Provide findings in these categories:

  • Critical: Issues that would cause the prompt to fail
  • Important: Issues that could cause problems or confusion
  • Suggestions: Improvements that aren't strictly necessary

For each finding, include:

  1. The specific location (section number and relevant text)
  2. What the issue is
  3. Suggested fix

Example output

code
## Critical Issues

### 1. Version mismatch (Section 2.5 vs 7.3)
- **Location**: Section 2.5 pins `next: "^14.0.0"`, Section 7.3 uses `create-next-app@15`
- **Issue**: These versions conflict - the installed version won't match the pinned version
- **Fix**: Align both to the same major version

## Important Issues

### 1. Missing step reference (Section 8.1)
- **Location**: "Use the repository name gathered in Section 7.0"
- **Issue**: Section 7.0 asks about repo name, but Section 8.1 also has AskUserQuestion for repo name
- **Fix**: Remove duplicate question from Section 8.1

## Suggestions

### 1. Consider adding timeout handling
- **Location**: Section 9.3 deployment validation
- **Issue**: No guidance on what to do if Vercel takes longer than 30 seconds
- **Fix**: Add polling logic or extended timeout instructions