/meta-reviewing-reviewing
Code review patterns, feedback principles. Use when reviewing PRs, implementations, or making approval/rejection decisions. Covers self-correction, progress tracking, feedback principles, severity levels.
$ npx -y skills add agents-inc/skills --skill meta-reviewing-reviewing --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/meta-reviewing-reviewing
Context preview
The summary Claude sees to decide when to auto-load this skill.
Code review patterns, feedback principles. Use when reviewing PRs, implementations, or making approval/rejection decisions. Covers self-correction, progress tracking, feedback principles, severity levels.
SKILL.md
meta-reviewing-reviewing.SKILL.mdname: meta-reviewing-reviewing
description: Code review patterns, feedback principles. Use when reviewing PRs, implementations, or making approval/rejection decisions. Covers self-correction, progress tracking, feedback principles, severity levels.
Reviewing Patterns
> **Quick Guide:** Read ALL files completely before commenting. Provide specific file:line references for every issue. Distinguish severity (Must Fix vs Should Fix vs Nice to Have). Explain WHY, not just WHAT. Suggest solutions following existing patterns. Acknowledge good work - positive reinforcement teaches what to repeat.
---
<critical_requirements>
CRITICAL: Before Any Review
**(You MUST read ALL files mentioned in the PR/spec completely before providing feedback)**
**(You MUST provide specific file:line references for every issue found)**
**(You MUST distinguish severity: Must Fix vs Should Fix vs Nice to Have)**
**(You MUST explain WHY something is an issue, not just WHAT is wrong)**
**(You MUST verify success criteria are met with evidence before approving)**
**(You MUST acknowledge what was done well - not just issues)**
</critical_requirements>
---
**Auto-detection:** code review, PR review, pull request, review code, check implementation, verify changes
**When to use:**
- Reviewing any code changes (PRs, implementations, specs)
- Providing structured feedback on code quality
- Making approval/rejection decisions
- Ensuring codebase standards are maintained
**When NOT to use:**
- When implementing code (use developer skills instead)
- For automated linting/type-checking (use CI/CD tooling)
- For deep security audits (use dedicated security review)
- For high-level architecture decisions (use planning/PM workflows)
**Key patterns covered:**
- Self-correction checkpoints for reviewers
- Post-action reflection after reviews
- Progress tracking for multi-file reviews
- Feedback principles (specific, explain why, suggest solutions, severity, acknowledge good)
- Decision framework for approval/rejection
- Review-specific anti-patterns (scope creep, refactoring, not using utilities)
**Detailed Resources:**
- [examples/core.md](examples/core.md) - All examples: progress tracking, feedback patterns, anti-patterns
---
<philosophy>
Philosophy
Code review is about **improving code quality while teaching good patterns**. Every piece of feedback should help the author become a better developer. Be direct but constructive.
**When reviewing code:**
- Always read the full context before commenting
- Base feedback on facts, not assumptions
- Distinguish blocking issues from improvements
- Teach through your feedback - explain the "why"
- Recognize good work to reinforce patterns
**When NOT to be harsh:**
- Don't nitpick style when code is functionally correct
- Don't request changes for personal preference
- Don't block PRs for minor issues that can be follow-ups
- Don't forget that the author worked hard on this
**Core principles:**
- **Evidence-based**: Base all feedback on what you actually read
- **Actionable**: Every issue should have a clear path to resolution
- **Proportional**: Match feedback severity to actual impact
- **Educational**: Help authors understand WHY, not just WHAT
- **Balanced**: Acknowledge good work alongside issues
</philosophy>
---
<patterns>
Core Patterns
Pattern 1: Self-Correction Triggers
These checkpoints prevent review drift and ensure thorough analysis. Check yourself throughout the review process.
**Self-Correction Checkpoints:**
| Trigger | Correction | | ------------------------------------------------- | ------------------------------------------ | | Providing feedback without reading full file | Stop. Read the complete file first. | | Saying "this needs improvement" without specifics | Stop. Provide file:line references. | | Approving without checking success criteria | Stop. Verify each criterion with evidence. | | Focusing only on issues | Stop. Add positive feedback. | | Making assumptions about code behavior | Stop. Read the actual implementation. | | Flagging issues without explaining WHY | Stop. Add rationale for each issue. | | Reviewing code outside your domain | Stop. Defer to specialist reviewer. |
---
Pattern 2: Post-Action Reflection
After completing your review, verify quality before finalizing.
**Reflection Questions:**
1. Did I read all relevant files completely before commenting? 2. Did I check against all success criteria in the spec? 3. Are my issues specific (file:line) and actionable? 4. Did I distinguish severity correctly (blocker vs improvement)? 5. Did I acknowledge what was done well? 6. Should any part go to a specialist reviewer? 7. Is my recommendation (approve/request changes) justified?
**Only finalize review when you can answer "yes" to all applicable questions.**
---
Pattern 3: Progress Tracking
For multi-file reviews, track your progress to maintain orientation.
**Track These Elements:**
1. **Files Examined:** [list of files read completely] 2. **Success Criteria Status:** [checked/unchecked for each criterion] 3. **Issues Found:** [categorized by severity] 4. **Positive Patterns Noted:** [what was done well] 5. **Deferred Items:** [what needs specialist review]
For tracking examples, see [examples/core.md](examples/core.md).
---
Pattern 4: Feedback Principles
All feedback should follow these principles for maximum effectiveness.
Be Specific
Every issue needs a precise location and actionable detail.
Explain Why
Don't just say what's wrong -- explain the impact so authors learn.
Suggest Solutions
Point to existing patterns when possible.
Distinguish Severity
Use clear markers to communicate priority:
| Marker | Category
Read more
name: meta-reviewing-reviewing description: Code review patterns, feedback principles. Use when reviewing PRs, implementations, or making approval/rejection decisions. Covers self-correction, progress tracking, feedback principles, severity levels.
Reviewing Patterns
> **Quick Guide:** Read ALL files completely before commenting. Provide specific file:line references for every issue. Distinguish severity (Must Fix vs Should Fix vs Nice to Have). Explain WHY, not just WHAT. Suggest solutions following existing patterns. Acknowledge good work - positive reinforcement teaches what to repeat.
---
<critical_requirements>
CRITICAL: Before Any Review
**(You MUST read ALL files mentioned in the PR/spec completely before providing feedback)**
**(You MUST provide specific file:line references for every issue found)**
**(You MUST distinguish severity: Must Fix vs Should Fix vs Nice to Have)**
**(You MUST explain WHY something is an issue, not just WHAT is wrong)**
**(You MUST verify success criteria are met with evidence before approving)**
**(You MUST acknowledge what was done well - not just issues)**
</critical_requirements>
---
**Auto-detection:** code review, PR review, pull request, review code, check implementation, verify changes
**When to use:**
- Reviewing any code changes (PRs, implementations, specs)
- Providing structured feedback on code quality
- Making approval/rejection decisions
- Ensuring codebase standards are maintained
**When NOT to use:**
- When implementing code (use developer skills instead)
- For automated linting/type-checking (use CI/CD tooling)
- For deep security audits (use dedicated security review)
- For high-level architecture decisions (use planning/PM workflows)
**Key patterns covered:**
- Self-correction checkpoints for reviewers
- Post-action reflection after reviews
- Progress tracking for multi-file reviews
- Feedback principles (specific, explain why, suggest solutions, severity, acknowledge good)
- Decision framework for approval/rejection
- Review-specific anti-patterns (scope creep, refactoring, not using utilities)
**Detailed Resources:**
- [examples/core.md](examples/core.md) - All examples: progress tracking, feedback patterns, anti-patterns
---
<philosophy>
Philosophy
Code review is about **improving code quality while teaching good patterns**. Every piece of feedback should help the author become a better developer. Be direct but constructive.
**When reviewing code:**
- Always read the full context before commenting
- Base feedback on facts, not assumptions
- Distinguish blocking issues from improvements
- Teach through your feedback - explain the "why"
- Recognize good work to reinforce patterns
**When NOT to be harsh:**
- Don't nitpick style when code is functionally correct
- Don't request changes for personal preference
- Don't block PRs for minor issues that can be follow-ups
- Don't forget that the author worked hard on this
**Core principles:**
- **Evidence-based**: Base all feedback on what you actually read
- **Actionable**: Every issue should have a clear path to resolution
- **Proportional**: Match feedback severity to actual impact
- **Educational**: Help authors understand WHY, not just WHAT
- **Balanced**: Acknowledge good work alongside issues
</philosophy>
---
<patterns>
Core Patterns
Pattern 1: Self-Correction Triggers
These checkpoints prevent review drift and ensure thorough analysis. Check yourself throughout the review process.
**Self-Correction Checkpoints:**
| Trigger | Correction | | ------------------------------------------------- | ------------------------------------------ | | Providing feedback without reading full file | Stop. Read the complete file first. | | Saying "this needs improvement" without specifics | Stop. Provide file:line references. | | Approving without checking success criteria | Stop. Verify each criterion with evidence. | | Focusing only on issues | Stop. Add positive feedback. | | Making assumptions about code behavior | Stop. Read the actual implementation. | | Flagging issues without explaining WHY | Stop. Add rationale for each issue. | | Reviewing code outside your domain | Stop. Defer to specialist reviewer. |
---
Pattern 2: Post-Action Reflection
After completing your review, verify quality before finalizing.
**Reflection Questions:**
1. Did I read all relevant files completely before commenting? 2. Did I check against all success criteria in the spec? 3. Are my issues specific (file:line) and actionable? 4. Did I distinguish severity correctly (blocker vs improvement)? 5. Did I acknowledge what was done well? 6. Should any part go to a specialist reviewer? 7. Is my recommendation (approve/request changes) justified?
**Only finalize review when you can answer "yes" to all applicable questions.**
---
Pattern 3: Progress Tracking
For multi-file reviews, track your progress to maintain orientation.
**Track These Elements:**
1. **Files Examined:** [list of files read completely] 2. **Success Criteria Status:** [checked/unchecked for each criterion] 3. **Issues Found:** [categorized by severity] 4. **Positive Patterns Noted:** [what was done well] 5. **Deferred Items:** [what needs specialist review]
For tracking examples, see [examples/core.md](examples/core.md).
---
Pattern 4: Feedback Principles
All feedback should follow these principles for maximum effectiveness.
Be Specific
Every issue needs a precise location and actionable detail.
Explain Why
Don't just say what's wrong -- explain the impact so authors learn.
Suggest Solutions
Point to existing patterns when possible.
Distinguish Severity
Use clear markers to communicate priority:
| Marker | Category
Showing the first part of this file.
The official skills marketplace for Agents Inc. 150+ skills covering everything from React and Prisma to Redis, ElevenLabs, and infrastructure tooling. Pick the skills that match your stack and install them via Claude Code. Need more control?
Repo: agents-inc/skills
Other skills on agents-inc-skills.
- /ai-infrastructure-huggingface-inference
Hugging Face Inference SDK patterns for TypeScript/Node.js — InferenceClient setup, chat completion, text generation, streaming, embeddings, image generation, audio transcription, translation, summarization, and Inference Endpoints
Open skill - /ai-infrastructure-litellm
LiteLLM proxy server setup, TypeScript client patterns via OpenAI SDK, model routing, fallbacks, load balancing, spend tracking, virtual keys, and production deployment
Open skill - /ai-infrastructure-modal
Serverless GPU compute platform for AI model deployment — web endpoints, GPU functions, model serving, and TypeScript client patterns
Open skill - /ai-infrastructure-ollama
Local LLM inference with the Ollama JavaScript client -- chat, streaming, tool calling, vision, embeddings, structured output, model management, and OpenAI-compatible endpoint
Open skill - /ai-infrastructure-replicate
Replicate SDK patterns for TypeScript/Node.js -- client setup, predictions, streaming, webhooks, file handling, model versioning, deployments, and training
Open skill - /ai-infrastructure-together-ai
Together AI SDK patterns for TypeScript — client setup, chat completions, streaming, structured output, function calling, embeddings, image generation, fine-tuning, and OpenAI-compatible endpoints
Open skill

