GitHub Plugin for Claude Code

GitHub automation plugin for CI/CD, pull requests, and code review workflows.

Features

• 🔍 Auto-detect CI context - Works with PRs, feature branches, and main/master
• 📊 Intelligent log parsing - Identifies lint, test, type, and build errors
• 🔧 Automated fixes - Applies targeted fixes for common CI failures
• 🛡️ Safety checks - Warns about main branch changes and uncommitted work
• 🎯 Specialized agents - Dedicated agents for log analysis and error fixing
• 📝 PR creation - Generate well-formatted pull requests from GitHub issues
• 💬 PR comment resolution - Interactive or autonomous review comment resolution
• 🤖 AI confidence scoring - Smart prioritization of review comments (0-100 score)
• 🔄 Rollback safety - Auto-stash changes in autonomous mode for easy undo

Installation

1. Copy this plugin to your Claude Code plugins directory:

   cp -r github-plugin ~/.claude/plugins/

2. Install dependencies:

   cd ~/.claude/plugins/github-plugin
   bun install

3. Restart Claude Code to load the plugin

Prerequisites

• GitHub CLI (gh) installed and authenticated
• Bun for TypeScript runtime
• Python projects: uv for package management
• TypeScript projects: Bun already installed

Commands

/fix-ci

Auto-detect, analyze, and fix CI/CD failures.

Usage:

/fix-ci              # Fix CI on current branch
/fix-ci 123          # Fix CI for PR #123
/fix-ci https://...  # Fix CI for PR URL

What it does:

1. Detects context: Identifies whether you're on a PR, feature branch, or main branch
2. Fetches logs: Downloads failed CI job logs using GitHub CLI
3. Analyzes errors: Uses ci-log-analyzer agent to parse logs and identify:
   • Lint/format errors (Ruff, ESLint)
   • Test failures (pytest, jest)
   • Type errors (mypy, TypeScript)
   • Build errors (syntax, imports)
4. Applies fixes: Uses ci-error-fixer agent to automatically fix issues
5. Reports results: Shows diffs and completion summary

Safety checks:

• Warns when operating on main/master branch
• Alerts about uncommitted changes
• Flags unclear fixes for manual review
• Asks about priorities when multiple failure types exist

/create-pr

Create GitHub Pull Request with proper formatting and SDLC best practices.

Usage:

/create-pr <issue_number>                    # Basic PR from issue
/create-pr <issue_number> path/to/plan.md    # PR with implementation plan

What it does:

1. Fetches issue details: Uses gh issue view to get issue info
2. Generates PR title: Format: <issue_type>: #<issue_number> - <issue_title>
   • Examples: feat: #123 - Add user authentication, bug: #456 - Fix login error
3. Creates PR body with SDLC best practices:
   • Summary: Explains the "why" and business context
   • Review Focus: Highlights architectural decisions, gotchas, and file review order
   • Testing Notes: Only for manual scenarios beyond CI automation
   • References: Links to issue, plan, and related documentation
   • Explicitly avoids redundant commit listings and generic testing documentation
4. Reviews changes: Shows git diff and commit history
5. Pushes and creates: Pushes branch and creates PR with gh pr create
6. Returns PR URL: Outputs only the PR URL for easy access

Example:

$ /create-pr 123 docs/implementation-plan.md

# Reviews changes, creates PR with best-practice formatting, returns:
https://github.com/user/repo/pull/456

Best Practices (Enforced in Instructions):

• ✅ Explains the "why" and business rationale
• ✅ Summarizes changes without listing every commit
• ✅ Highlights review focus: files, architecture decisions, gotchas
• ✅ Includes testing notes only for non-obvious scenarios
• ❌ No redundant commit sections (visible in PR's Commits tab)
• ❌ No generic testing documentation (CI/CD pipeline is authoritative)
• ❌ No boilerplate or placeholder sections

See research/pr-description-best-practices.md for research-backed guidance and authoritative sources.

/create-issue-from-plan

Convert implementation plans into GitHub Issues with automatic checkboxes and linking.

Usage:

/create-issue-from-plan plans/feature-name.md

What it does:

1. Parses plan file: Reads YAML frontmatter and markdown structure
2. Extracts content:
   • Overview section → Issue body summary
   • Implementation Phases → Checklist items
   • Validation Commands → Issue section
   • Acceptance Criteria → Issue checklist
   • Research files → Linked in Issue body
3. Creates Issue: Generates GitHub Issue with:
   • Title: <type>: <plan title>
   • Body: Complete implementation checklist with all phases
   • Labels: status:planning + type label (feat, bug, chore, etc.)
4. Updates plan file: Adds Issue number to plan frontmatter (issue: #123)
5. Commits linkage: Saves Issue reference in plan file via git commit

Plan Frontmatter:

---
title: "Feature Description"
type: Feature  # Bug|Feature|Chore|Refactor|Enhancement|Documentation
issue: null    # Auto-populated with Issue #123 after creation
research:
  - research/related-research.md
status: Draft
created: 2024-11-16
---

Example:

$ /create-issue-from-plan plans/add-oauth2-auth.md

✅ Issue #456 created from plan
📝 Updated plan frontmatter: issue: 456
🔗 Linked research files in Issue body
💾 Committed Issue linkage to plan file

Issue: https://github.com/user/repo/issues/456

/address-pr-comments

Interactive or autonomous workflow for addressing PR review comments with AI-powered confidence scoring.

Modes:

The command automatically detects the execution environment:

• Interactive Mode (terminal): User selects which comments to address
• Autonomous Mode (CI/CD): Auto-addresses high-confidence comments (score ≥ 80)

Usage:

# Interactive mode (default in terminal)
/address-pr-comments 18
/address-pr-comments https://github.com/owner/repo/pull/123

# Autonomous mode (auto-detected in CI/CD)
/address-pr-comments 18 --autonomous
/address-pr-comments 18 auto

# Force interactive mode
/address-pr-comments 18 --interactive

What it does:

1. Fetches PR data: Downloads PR details, review comments, and issue comments using gh CLI
2. Analyzes & scores comments: Filters actionable comments, categorizes by type, and assigns confidence scores (0-100)
3. Presents or filters: Interactive shows all comments; Autonomous filters for high-confidence (≥80)
4. Applies changes: Reads files, explains changes, and uses Edit tool to apply fixes
5. Reports results: Shows detailed summary with addressed items and skipped items

Confidence Scoring (0-100):

Comments are scored based on:

• Specificity (30pts): File path + line number + concrete suggestion
• Language clarity (25pts): Directive ("must", "please") vs suggestive ("consider")
• Category (25pts): STYLE (25) > DOCS (20) > CODE (15) > TEST (10)
• Reviewer authority (10pts): Maintainer/owner (10) > contributor (3)
• Discussion status (10pts): Blocking (10) > unresolved (7)

Thresholds:

• 80-100 (High): Auto-addressed in autonomous mode
• 60-79 (Medium): Presented in interactive mode
• 0-59 (Low): Skipped or flagged for manual review

Comment categories:

• CODE: Logic changes, bug fixes, refactoring
• STYLE: Formatting, naming conventions
• DOCS: Documentation improvements
• TEST: Test coverage, test improvements
• QUESTIONS: Clarifications needed

Safety features:

• Autonomous mode: Creates git stash rollback point before changes
• No automatic commits or pushes
• Shows changes and confidence scores
• Skips ambiguous or low-confidence comments
• Detailed reporting of what was/wasn't addressed

Interactive Mode Example:

$ /address-pr-comments 42

Found 3 comments on PR #42: Add user authentication

1. [CODE] src/auth.ts:25 (Score: 90)
   Comment: "Add null check before accessing user.email"

2. [STYLE] src/utils.ts:42 (Score: 95)
   Comment: "Use const instead of let"

Which items would you like me to address? all

✓ Addressed item 1: Added null check
✓ Addressed item 2: Changed let to const

📊 Summary: 2 comments addressed, 2 files modified

Autonomous Mode Example:

$ /address-pr-comments 42 --auto

🤖 AUTONOMOUS MODE ACTIVE

Analyzing 5 comments on PR #42
Confidence threshold: 80/100

High-Confidence Items (2):
✓ [STYLE] src/utils.ts:42 (95) - Changed let to const
✓ [CODE] src/auth.ts:25 (90) - Added null check

Skipped Items (3):
⏭️ [CODE] src/api.ts:55 (65) - Vague suggestion
⏭️ [TEST] tests/auth.test.ts (45) - Question without action
⏭️ [QUESTIONS] src/models.ts (35) - Ambiguous feedback

🔄 Rollback: git stash apply stash^{/pre-autonomous-pr-42-...}
📊 Summary: 2 addressed, 3 skipped, 2 files modified

/review-pr

Comprehensive PR review with detailed analysis and actionable feedback.

Usage:

/review-pr 123                              # Review PR by number
/review-pr https://github.com/owner/repo/pull/456  # Review by URL
/review-pr owner/repo#789                   # Review with repo context

What it does:

1. Fetches PR data: Downloads PR details, diff, comments, and CI status using gh CLI
2. Checks out branch: Switches to PR branch for local analysis
3. Analyzes changes: Reviews all modified files, commit history, and diff
4. Reviews discussions: Examines comments, unresolved threads, and CI results
5. Generates report: Provides structured review with executive summary, code quality analysis, file-by-file feedback

Review output includes:

• Executive Summary: Purpose, scope, risk level, recommendation (Approve/Request Changes)
• Code Quality Analysis: Architecture, style, testing, documentation, error handling, security
• File-by-File Review: Detailed feedback for each changed file
• Discussion Review: Unresolved conversations, CI status, action items
• Recommendations: Must fix, should fix, consider, and praise items

Safety checks:

• Warns about uncommitted changes before checkout
• Requires GitHub CLI authentication
• Respects repository permissions
• Checks project conventions (CONTRIBUTING.md, PR templates)

Agents

ci-log-analyzer

Specialized agent for parsing CI logs and identifying error patterns.

Capabilities:

• Recognizes 10+ error pattern types
• Extracts file paths, line numbers, and error messages
• Categorizes by type and severity
• Returns structured JSON for automated fixing

Tools: Read, Grep, Bash

ci-error-fixer

Specialized agent for applying fixes based on error types.

Capabilities:

• Applies targeted fixes for each error category
• Shows diffs for all changes
• Flags complex issues for manual review
• Respects code style and best practices

Tools: Read, Edit, Write, Bash

Supported Error Types

Category	Examples	Fix Strategy
Lint	Ruff formatting, unused imports, ESLint errors	Auto-format, remove unused code
Test	pytest/jest failures, assertion errors, missing imports	Fix logic, add imports, update tests
Type	mypy/TypeScript type mismatches, missing returns	Add type hints, fix type errors
Build	Syntax errors, import errors, compilation failures	Fix syntax, resolve imports

Example Workflow

# You're on a feature branch with failing CI
$ /fix-ci

# Plugin detects context
✓ Detected feature branch: feature/user-auth
✓ Found PR #42: "Add user authentication"
✓ Identified 3 failed checks

# Fetches and analyzes logs
✓ Analyzing CI logs...
  Found 5 errors: 2 lint, 2 test, 1 type

# Applies fixes
✅ Fixed: src/auth.py:15 - Removed unused import 'os'
✅ Fixed: src/auth.py:42 - Added missing type hint
✅ Fixed: tests/test_auth.py:10 - Added missing import 'requests'
⚠️  Flagged: tests/test_auth.py:55 - Assertion failure needs review

# Summary
🎉 Fix Summary
  ✅ 3 issues fixed
  ⚠️  1 issue flagged for manual review

Next steps:
  1. Review changes: git diff
  2. Run tests locally: pytest
  3. Commit: git add . && git commit -m "fix: CI failures"
  4. Push: git push

Development

Validate Plugin

bun run validate

Run Tests

# Test plugin manifest validation
bun run validate

# Test in Claude Code
# Copy to plugins directory and restart Claude Code

Contributing

1. Make changes to commands or agents
2. Run validation: bun run validate
3. Test in Claude Code
4. Update CHANGELOG.md
5. Submit PR

Project Structure

github-plugin/
├── .claude-plugin/
│   └── plugin.json              # Plugin manifest
├── .github/workflows/
│   └── validate-plugin.yml      # CI validation
├── commands/
│   ├── fix-ci.md                # CI fixing command
│   ├── create-pr.md             # PR creation command
│   ├── address-pr-comments.md   # PR comment resolution
│   └── review-pr.md             # PR review command
├── agents/
│   ├── ci-log-analyzer.md       # Log parsing agent
│   └── ci-error-fixer.md        # Error fixing agent
├── scripts/
│   └── validate-plugin.ts       # Validation script
├── package.json
├── README.md
├── CHANGELOG.md
└── LICENSE

Troubleshooting

"gh: command not found"

Install GitHub CLI:

# macOS
brew install gh

# Linux
sudo apt install gh

# Windows
winget install GitHub.cli

"Not authenticated with GitHub"

Authenticate:

gh auth login

"No failed CI runs found"

Check that CI has actually run:

gh run list --limit 5

Plugin not loading

1. Verify plugin is in ~/.claude/plugins/github-plugin/
2. Check plugin.json is valid: bun run validate
3. Restart Claude Code

License

MIT License - see LICENSE file for details

Author

Ladislav Martincik (@iamladi)

Links

• Claude Code Documentation
• GitHub CLI Documentation
• Plugin Repository