openclaw-github-pr-reviewer

GitHub PR Reviewer with aimemo Memory

A complete example OpenClaw skill that demonstrates aimemo integration for persistent memory across sessions.

What This Skill Does

Reviews GitHub Pull Requests and learns your code style preferences over time:

• 🧠 Remembers your naming conventions, formatting rules, and patterns
• 📝 Stores new discoveries from each review session
• 🔄 Improves with every PR reviewed
• 🎯 Applies learned preferences consistently

Prerequisites

1. aimemo installed (v0.4.0+):

   # Linux/macOS
   curl -sSL https://raw.githubusercontent.com/MyAgentHubs/aimemo/main/install.sh | bash
   
   # Or macOS Homebrew
   brew install MyAgentHubs/tap/aimemo

2. aimemo MCP server registered with OpenClaw:

   claude mcp add-json aimemo-memory '{"command":"aimemo","args":["serve"]}'

3. Workspace memory initialized:

   cd ~/.openclaw/workspace
   aimemo init

4. GitHub token (for API access):

   export GITHUB_TOKEN="ghp_your_token_here"

Installation

Option 1: Copy to OpenClaw Skills Directory

# Copy this skill to your OpenClaw workspace
cp -r examples/openclaw-github-pr-reviewer \
  ~/.openclaw/workspace/skills/

Option 2: Symlink (for development)

# Symlink for easier updates
ln -s $(pwd)/examples/openclaw-github-pr-reviewer \
  ~/.openclaw/workspace/skills/github-pr-reviewer-aimemo

Option 3: Install from ClawHub (Coming Soon)

claude skill install github-pr-reviewer-aimemo

Usage

First Review (Learning Phase)

You: "Review PR #123 from my project"

Skill:

1. Asks about your basic preferences:
   • "Do you prefer snake_case or camelCase?"
   • "How do you handle errors?"
   • "What's your indentation style?"
2. Reviews the PR
3. Stores your answers for future reviews

Subsequent Reviews (Using Learned Preferences)

You: "Review PR #456"

Skill:

1. Automatically loads learned preferences
2. Reviews PR using your established rules
3. Flags issues based on your standards
4. Stores any new patterns discovered

Correcting Mistakes

You: "Actually, I'm fine with trailing commas in arrays"

Skill:

1. Updates preference in memory
2. Acknowledges: "Got it, I'll remember that"
3. Applies corrected rule in future reviews

Example Session

You: Review https://github.com/myorg/myproject/pull/789

Skill: [Loads memory: "User prefers snake_case, explicit error types, 4-space indent"]

Skill: I've reviewed PR #789. Here's my feedback:

## Summary
Request changes - found 3 style issues

## Key Issues
1. Function `getUserData` should be `get_user_data` (snake_case preference)
2. Error on line 45 should have explicit type (AppError expected)
3. Indentation inconsistent (mix of 2 and 4 spaces)

## Suggestions
- Consider adding tests for the new validation logic
- The error message on line 67 could be more descriptive

## Positive Notes
- Good use of early returns
- Clear variable names
- Well-documented function

[Stores: "User wants consistent 4-space indent", "Validation logic needs tests"]

You: Great! Also, for future reviews, I want you to always check if the changelog was updated.

Skill: Understood! I've added "Verify CHANGELOG.md updated" to my review checklist. I'll check this in all future PR reviews.

[Stores: "Must verify CHANGELOG.md updated" in review-checklist]

What Gets Stored

Code Style Preferences

• Naming conventions (snake_case, camelCase, PascalCase)
• Indentation (spaces/tabs, 2/4 spaces)
• Formatting (trailing commas, semicolons, line breaks)

Patterns & Best Practices

• Error handling approaches
• File organization patterns
• Testing standards
• Documentation requirements

Anti-Patterns to Flag

• Specific issues user wants avoided
• Project-specific no-nos
• Security concerns

User Corrections

• When you correct the skill's feedback
• Clarifications on preferences
• Exception cases

Project Context

• Tech stack info
• Team conventions
• Domain-specific rules

Memory Structure

All data is stored in:

~/.openclaw/workspace/.aimemo/memory-github-pr-reviewer-aimemo.db

Entities:

• code-style-preferences: Naming, formatting rules
• error-handling-patterns: How to handle errors
• common-mistakes: Things to always flag
• user-corrections: Feedback from user
• project-context: Project-specific info
• review-checklist: Things to always check

Journal entries: Session summaries showing learning progress

Inspecting Memory

List all stored preferences

aimemo list --context github-pr-reviewer-aimemo

Search for specific topic

aimemo search "error handling" --context github-pr-reviewer-aimemo

Get specific entity

aimemo get code-style-preferences --context github-pr-reviewer-aimemo

Export everything

aimemo export --context github-pr-reviewer-aimemo --format json > memory.json

Read journal entries

aimemo search --journal --since 7d --context github-pr-reviewer-aimemo

Customization

Modify SKILL.md

Edit the instructions to:

• Add language-specific rules (Python, Go, TypeScript, etc.)
• Include framework-specific checks (React, Django, etc.)
• Add CI/CD integration checks
• Customize review format

Add Custom Entity Types

In SKILL.md instructions:

Store security-specific rules:
memory_store({
  context: "github-pr-reviewer-aimemo",
  entities: [{
    name: "security-rules",
    entityType: "security",  // Custom type
    observations: ["Never commit .env files"],
    tags: ["security", "critical"]
  }]
})

Troubleshooting

Memory not loading

Check if database exists:

ls -lh ~/.openclaw/workspace/.aimemo/memory-github-pr-reviewer-aimemo.db

Verify context parameter in SKILL.md (must be exact):

{context: "github-pr-reviewer-aimemo"}

Skill forgets preferences

Ensure memory_store is called:

• Check OpenClaw logs for successful store operations
• Verify database is writable: touch ~/.openclaw/workspace/.aimemo/test && rm ~/.openclaw/workspace/.aimemo/test

Wrong preferences loaded

Update incorrect preference:

# Via CLI
aimemo observe code-style-preferences "CORRECTION: Use camelCase not snake_case" \
  --context github-pr-reviewer-aimemo

# Or let skill overwrite in next session

Database corruption

Recover (rare):

cd ~/.openclaw/workspace/.aimemo
sqlite3 memory-github-pr-reviewer-aimemo.db ".recover" | \
  sqlite3 memory-recovered.db
mv memory-github-pr-reviewer-aimemo.db memory-backup.db
mv memory-recovered.db memory-github-pr-reviewer-aimemo.db

Start fresh:

rm ~/.openclaw/workspace/.aimemo/memory-github-pr-reviewer-aimemo.db
# Skill will auto-create on next use

Advanced Usage

Multiple Projects

Use different skill instances per project:

# Install multiple times with different names
cp SKILL.md ~/.openclaw/workspace/skills/pr-reviewer-project-a/SKILL.md
cp SKILL.md ~/.openclaw/workspace/skills/pr-reviewer-project-b/SKILL.md

# Each gets isolated memory:
# memory-pr-reviewer-project-a.db
# memory-pr-reviewer-project-b.db

Team Shared Preferences

Export and share baseline preferences:

# Export your preferences
aimemo export --context github-pr-reviewer-aimemo --format json > team-standards.json

# Team member imports
aimemo import team-standards.json --context github-pr-reviewer-aimemo

Periodic Review

Check what skill has learned:

# Weekly review
aimemo list --context github-pr-reviewer-aimemo | less

# See recent changes
aimemo search --journal --since 7d --context github-pr-reviewer-aimemo

Learning Progress Example

After 1 Review

code-style-preferences:
  - Prefer snake_case for variables
  - 4-space indentation

After 5 Reviews

code-style-preferences:
  - Prefer snake_case for variables
  - 4-space indentation
  - No trailing commas in objects
  - PascalCase for classes

error-handling-patterns:
  - Always use explicit error types (AppError)
  - Wrap third-party errors with context

After 20 Reviews

code-style-preferences: 12 observations
error-handling-patterns: 8 observations
common-mistakes: 15 observations
review-checklist: 7 items
project-context: 10 observations

Journal entries: 20 sessions documented

The skill becomes increasingly valuable as its knowledge base grows.

Contributing

Found an improvement? Submit a PR to: https://github.com/MyAgentHubs/aimemo

Ideas for enhancements:

• Language-specific templates
• Integration with linters (ESLint, Pylint, etc.)
• Auto-formatting suggestions
• Security scanning integration
• Performance analysis

License

MIT - See LICENSE

See Also

• aimemo Documentation
• OpenClaw Integration Guide
• OpenClaw Workflow Details