🐙 GitHub ディープ統合

コードコラボレーション · リポジトリ管理 · DevOps

Turn the world's largest code hosting platform into your AI agent's command center! Grant LLMs full-stack capabilities to browse repository code, auto-create Issues and Pull Requests, manage CI/CD pipelines, and perform code reviews.

OpenClaw チーム

🚀 クイックインストール

ターミナルで以下のコマンドを実行してインストール：

npx clawhub install github

📊 統計概要

🎛️ 仕組み

This skill is the core bridge connecting AI to software engineering infrastructure within the OpenClaw ecosystem. After integrating with GitHub's REST & GraphQL APIs, the LLM gains the ability to intervene across the entire development lifecycle:

• 📂 Deep Repository Traversal: The model can traverse all repository structures under an organization, read specific file contents by path, parse package.json, trace Git commit history and branch diff graphs — achieving holographic awareness of the project codebase.
• 🐛 Smart Issue Creation & Triage: When AI discovers potential bugs or improvements during conversation, it can directly create well-structured GitHub Issues with automatic label assignment (bug, enhancement, priority:high), eliminating tedious manual transcription.
• 🔀 Pull Request Automation Pipeline: The Agent can not only submit code changes to create PRs, but also attach comprehensive change summaries, link related Issue numbers, and even auto-request reviews from specific team members.
• ✅ CI/CD Pipeline Monitoring: Real-time tracking of GitHub Actions workflow run statuses. When builds fail, the model can directly pull failure step logs for analysis and provide fix recommendations.
• 👀 Code Review Assistance: Reads PR diff changes and performs line-by-line annotated review comments based on contextual understanding of code quality, security vulnerabilities, and performance concerns.

🧭 典型的なユースケース

🤖 シナリオ 1: Automated Bug Capture & Issue Filing

When AI catches an unexpected crash stack during automated test pipelines, the model immediately analyzes the error context and auto-creates a well-formatted Issue in the corresponding repository — complete with reproduction steps, error logs, and environment info, tagged as bug + critical.

📝 シナリオ 2: Intelligent Code Review Assistant

After team members submit PRs, AI automatically receives webhook notifications. It reads the diff file by file, cross-references the project's coding standards (from .eslintrc or CONTRIBUTING.md), and leaves constructive review comments on each suspicious change.

🚀 シナリオ 3: Release Automation Orchestration

When all CI checks pass on the main branch, the Agent can automatically create Release Tags, generate Changelogs (based on merged PR titles), trigger deployment pipelines, and push release notifications to Slack channels.

💻 コマンドリファレンス

インストール後、会話を通じてAIに自律的に呼び出させるか、CLIから手動で操作をトリガーできます：

View recent commits for a specified repository:

clawhub execute github action="commits" repo="openclaw/core" limit=10

Create an Issue with labels:

clawhub execute github action="create-issue" repo="openclaw/core" \
  title="Fix: Memory leak in streaming module" \
  labels="bug,priority:high" \
  body="Detected during automated stress test..."

Read PR diff changes and perform code review:

clawhub execute github action="review-pr" repo="openclaw/core" pr=42 \
  intent="Check for security vulnerabilities and performance issues"

Query recent GitHub Actions workflow run statuses:

clawhub execute github action="workflows" repo="openclaw/core" status="failure"

🛡️ 要件と認証

• 🔑 Personal Access Token: Go to GitHub Settings > Developer settings > Personal access tokens to create a fine-grained token with repo, workflow, and read:org permissions.
• ⚙️ Environment Variable Setup: Pre-set in terminal: export GITHUB_TOKEN="ghp_xxxxxxxxxxxx".
• 🔒 Security Recommendation: Never store tokens in plaintext in public repository .env files. Use GitHub Secrets or local encrypted key management solutions.

🔗 GitHubでソースを見る

© 2026 OpenClaw. All rights reserved.