Agent skill

plan-review

Iterative review of implementation plans

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/plan-review-charly-vibes-wai-2

SKILL.md

Iterative Plan Review (Rule of 5)

Perform thorough implementation plan review using the Rule of 5 - iterative refinement until convergence.

Setup

If plan path provided: Read the plan completely

If no plan path: Ask for the plan path or list available plans from plans/

Process

Perform 5 passes, each focusing on different aspects. After each pass (starting with pass 2), check for convergence.

PASS 1 - Feasibility & Risk

Focus on:

  • Technical feasibility of proposed changes
  • Identified risks and their mitigations
  • Dependencies on external factors (APIs, services, libraries)
  • Assumptions that need validation
  • Potential blockers not addressed
  • Resource requirements (time, expertise, infrastructure)

Output format:

PASS 1: Feasibility & Risk

Issues Found:

[FEAS-001] [CRITICAL|HIGH|MEDIUM|LOW] - Phase/Section
Description: [What's wrong]
Evidence: [Why this is a problem]
Recommendation: [How to fix with specific guidance]

[FEAS-002] ...

What to look for:

  • "We'll just..." statements hiding complexity
  • External dependencies assumed to be available
  • Time estimates that seem unrealistic
  • Required expertise not accounted for
  • Breaking changes not acknowledged
  • Rollback strategy missing or inadequate

PASS 2 - Completeness & Scope

Focus on:

  • Missing phases or steps
  • Undefined or vague success criteria
  • Gaps between current and desired state
  • "Out of Scope" clearly defined
  • All affected files/components identified
  • Testing strategy covers all scenarios
  • Edge cases considered

Prefix: COMP-001, COMP-002, etc.

What to look for:

  • "And then..." without the "then" being in the plan
  • Success criteria that can't be verified
  • Missing integration points
  • Unclear handoff points between phases
  • Scope creep hiding in vague language

PASS 3 - Spec & TDD Alignment

Focus on:

  • Links to spec files (if applicable)
  • Tests planned before implementation (TDD)
  • Success criteria are testable
  • Test-first approach clear in each phase
  • All requirements from specs covered
  • Verification steps defined

Prefix: TDD-001, TDD-002, etc.

What to look for:

  • "We'll test it" without specific test descriptions
  • Implementation before tests in phase sequence
  • Untestable success criteria ("works well", "is fast")
  • Missing test coverage for error cases
  • No mention of what tests to write

PASS 4 - Ordering & Dependencies

Focus on:

  • Phases in correct order (can't do B before A)
  • Dependencies between phases clearly stated
  • Parallelizable work identified
  • Critical path identified
  • Each phase can be independently verified
  • Rollback/reversal possible at phase boundaries

Prefix: ORD-001, ORD-002, etc.

What to look for:

  • Phase 2 requires Phase 3 to be done first
  • Circular dependencies between phases
  • "Big bang" integration at the end
  • No incremental verification
  • All-or-nothing approach

PASS 5 - Clarity & Executability

Focus on:

  • Specific enough for someone else to implement
  • File paths and changes are concrete
  • No ambiguous instructions
  • Clear what "done" means for each phase
  • Technical terms defined or understood
  • Handoff points between phases clear

Prefix: EXEC-001, EXEC-002, etc.

What to look for:

  • "Update the authentication" - update how? where?
  • "Make it work with..." - what does "work" mean?
  • "Refactor as needed" - refactor what? when?
  • File paths missing or vague
  • Assumptions about shared knowledge

Convergence Check

After each pass (starting with pass 2), report:

Convergence Check After Pass [N]:

1. New CRITICAL issues: [count]
2. Total new issues this pass: [count]
3. Total new issues previous pass: [count]
4. Estimated false positive rate: [percentage]

Status: [CONVERGED | ITERATE | NEEDS_HUMAN]

Convergence criteria:

  • CONVERGED: No new CRITICAL, <10% new issues vs previous pass, <20% false positives
  • ITERATE: Continue to next pass
  • NEEDS_HUMAN: Found blocking issues requiring human judgment

If CONVERGED before Pass 5: Stop and report final findings.

Final Report

After convergence or completing all passes:

## Plan Review Final Report

**Plan:** plans/[filename].md

### Summary

Total Issues by Severity:
- CRITICAL: [count] - Must fix before implementation
- HIGH: [count] - Should fix before implementation
- MEDIUM: [count] - Consider addressing
- LOW: [count] - Nice to have

Convergence: Pass [N]

### Top 3 Most Critical Findings

1. [FEAS-001] [Description] - Phase [N]
   Impact: [Why this matters]
   Fix: [What to do]

2. [COMP-003] [Description] - Phase [N]
   Impact: [Why this matters]
   Fix: [What to do]

3. [TDD-002] [Description] - Phase [N]
   Impact: [Why this matters]
   Fix: [What to do]

### Recommended Next Actions

1. [Action 1 - specific and actionable]
2. [Action 2 - specific and actionable]
3. [Action 3 - specific and actionable]

### Verdict

[READY_TO_IMPLEMENT | NEEDS_REVISION | NEEDS_MORE_RESEARCH]

**Rationale:** [1-2 sentences explaining the verdict]

Rules

  1. Be specific - Reference plan sections/phases, provide file:line if relevant
  2. Provide actionable fixes - Don't just say "add more detail", say what detail
  3. Validate claims - Don't flag potential issues, confirm they exist
  4. Prioritize correctly:
    • CRITICAL: Blocks implementation or will cause failure
    • HIGH: Significantly impacts quality or feasibility
    • MEDIUM: Worth addressing but not blocking
    • LOW: Minor improvements
  5. If converged before pass 5 - Stop and report, don't continue needlessly

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results