An automated AI agent skill that executes a rigorous, five-lens implementation plan review before executing code modifications.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install review-plan
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install review-plan using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Review Plan?
The review-plan skill is a powerful diagnostic tool designed for Openclaw Skills to safeguard codebases against premature or faulty implementations. It acts as an automated gatekeeper, analyzing proposed developer plans—such as those created by planning skills—for architectural integrity, security gaps, and adherence to Test-Driven Development (TDD) principles before writing any code.
By executing strict anti-confabulation protocols and loading technology-specific context on-demand, this skill ensures that plans align with the reality of the existing codebase. It acts as a safety layer for AI-driven development, providing developers and autonomous agents with a clear verdict, complete with copy-pasteable code fixes, to run before executing plan steps.
Review Plan Use Cases
- Pre-execution Plan Validation: Run automated reviews on new feature proposals to detect code mismatches and missing error handling before starting implementation.
- Orchestration Analysis: Evaluate complex plans to see if sub-tasks can be safely run in parallel by multi-agent subagents without circular dependencies.
- TDD Enforcement: Ensure that tasks systematically integrate failing test (RED) stages and minimal passing code (GREEN) stages without introducing defensive over-engineering.
How Review Plan Works
- Anti-Confabulation Verification: Emits a precise echo of the targeted plan step or file reference to confirm context alignment before making a judgment.
- Plan Parsing: Reads the plan file to extract core goals, architectural specifications, and tech stack declarations.
- Context-Specific Skill Loading: Automatically maps and loads relevant Openclaw Skills based on the identified tech stack (e.g., Python, React Router, Go, Vitest).
- Multi-Lens Execution: Runs five parallel or sequential reviews: Parallelization, TDD/Over-Engineering, Type & API verification, Library Best Practices, and Security/Edge Cases.
- Report Synthesis & Action: Compiles a Markdown report categorized by critical, major, and minor issues, saves it alongside the plan, and prompts the user for action.
Review Plan Setup
Configure the skill within your agent environment. Invoke the skill by providing the file path of your implementation plan:
# Example argument structure
Path: "docs/plans/2025-01-15-auth-feature.md"
Ensure that dependent code-review skills (such as python-code-review or go-code-review) are correctly pathed in your workspace to enable automatic skill loading in Step 2.
Review Plan Data Schema & Taxonomy
Review Artifact Location
The skill outputs a structured review file alongside the original plan:
- Original Plan:
[plan-dir]/[plan-name].md - Review File:
[plan-dir]/[plan-name]-review.md
Report Metadata Schema
| Section | Data Fields | Description |
|---|---|---|
| Header | Reviewed, Verdict |
Stores review execution timestamp and actionability verdict (Yes, With fixes, No). |
| Summary Table | Criterion, Status, Notes |
Evaluates adherence to the 5 lenses (Parallelization, TDD, Types/APIs, Libraries, Security). |
| Issues Found | Severity, Issue Code, Fix |
Lists bugs with copyable replacement blocks classified by critical, major, or minor severity. |
| Verdict & Reason | Verdict, Reasoning |
A brief summary explaining the execution readiness of the plan. |
Review Plan Advanced Features
- Zero-Gate Anti-Confabulation: Guarantees factual accuracy by requiring the AI to echo exact codebase lines or plan text before submitting verdicts.
- Multi-Agent Subagent Parallelization: Dynamically dispatches review lenses to distinct parallel subagents when the host platform supports concurrency.
- Dynamic Skill Chaining: Detects active file types (.py, .ts, .go) and frameworks to automatically load secondary Openclaw Skills for deeper technical checks.
SKILL.md
Loading
Related Openclaw Skills
Prompt Improver
An AI-agent tool designed to automatically audit and transform vague engineering prompts into clear, highly optimized instructions for code implementation, debugging, refactoring, and testing.
Remix v2 Data Flow Review
An advanced automated code review skill for analyzing and optimizing Remix v2 loader, action, and revalidation data flows.
respond-pr-feedback
An automated AI agent skill that posts contextual replies to GitHub Pull Request review comments and handles conversation thread resolution.
LLM Artifacts Review
Automatically audit and prune LLM-generated code artifacts, over-abstractions, and redundant tests across your codebase.
Release Notes Generator
An automated AI agent skill that analyzes Git histories, extracts merged pull requests, and generates strict Keep a Changelog compliant documentation.
Receive Feedback
A highly disciplined Openclaw Skills extension to automatically verify, organize, and execute fixes for code review feedback.
Featured*
