superpowers-requesting-code-review

What This Skill Does

The superpowers-requesting-code-review skill is a specialized agentic workflow designed for OpenClaw to enforce high-quality standards during software development. Unlike standard prompts, this skill leverages the sessions_spawn model to dispatch a dedicated code-reviewer subagent. By isolating the reviewer from your main session history, it ensures an objective, context-aware analysis of your work. The reviewer evaluates your git diffs, implementation details, and project requirements to catch bugs, design flaws, and specification gaps before they propagate into your codebase.

Installation

To integrate this skill into your environment, run the following command in your terminal: clawhub install openclaw/skills/skills/axelhu/openclaw-requesting-code-review

Use Cases

• Post-Task Validation: Automatically verify every subagent-dev task completion to ensure code correctness.
• Feature Gating: Perform mandatory reviews before merging significant features into the main branch.
• Refactoring Baseline: Run an analysis before major refactors to ensure existing logic is understood and protected.
• Debugging Complex Issues: Use the subagent as a second pair of eyes when stuck on elusive bugs to gain a fresh perspective on edge cases.
• Specification Compliance: Ensure your implementation aligns strictly with defined project requirements and prevents unnecessary scope creep.

Example Prompts

1. "I've just finished implementing the OAuth2 authentication module. Please dispatch the code reviewer subagent, using the diff at /tmp/auth-module.patch to verify compliance with the security requirements."
2. "Before I merge this refactor of the database layer, run a code review. Compare the current implementation against the original design document provided in the project root."
3. "The current task is complete, but I am unsure about the test coverage. Spawn a review subagent to identify any missing edge cases and rate the overall code quality."

Tips & Limitations

• Context is Key: Always provide the reviewer with specific git SHAs or patch files. The subagent performs best when it knows exactly what to analyze.
• Severity Matters: Treat 'Critical' findings as blockers. Do not proceed to the next development task until 'Important' issues are addressed.
• Technical Pushback: Do not blindly accept every suggestion. If you believe the reviewer has misunderstood the implementation, provide technical evidence and test results to refute the feedback; this is a collaborative tool, not a dictator.
• Efficiency: Limit reviews to meaningful chunks of work to keep the reviewer's focus sharp and ensure the sessions_spawn environment remains performant.