GitHub Autopilot

The Problem

Your GitHub notifications are a graveyard. PRs sit unreviewed for days. Bugs get filed and forgotten. CI breaks and nobody notices until the next deploy. You're spending engineering time on process instead of product — and it's killing your velocity.

What You Get

• Automated PR reviews — Reads diffs, identifies potential issues (security vulnerabilities, performance regressions, style violations, missing tests), and posts detailed review comments. Not "LGTM" — actual code review with specific line-by-line feedback.
• Bug triage and prioritization — Reads new issues, categorizes by severity and component, assigns priority labels, and suggests which engineer should look at it. Reduces triage meetings from 30 minutes to zero.
• Code generation — Writes implementation code, unit tests, and documentation for well-defined tasks. Understands your codebase patterns and follows existing conventions. Creates PRs with proper descriptions and linked issues.
• CI/CD monitoring — Watches build pipelines for failures, identifies the breaking commit, and notifies the responsible developer. Distinguishes between flaky tests and real failures.
• Sprint management — Tracks issue velocity, identifies blockers, and updates project boards. Knows when a sprint is at risk and surfaces it before standup.
• Dependency monitoring — Flags outdated dependencies, security advisories, and breaking changes in upcoming major versions. Prioritizes by actual risk, not just version number.

What Makes This Different From a Generic Prompt

• Codebase awareness — Understands your repo structure, naming conventions, and architectural patterns. Reviews code in context, not in isolation.
• Continuous monitoring — Doesn't wait to be asked. Watches for new PRs, issues, and CI failures and acts on them proactively.
• Git workflow native — Understands branching strategies, merge conflicts, rebase vs. merge, and conventional commits. Works within your existing workflow, doesn't impose a new one.
• Security-conscious — Scans for common vulnerabilities (SQL injection, XSS, exposed secrets, insecure dependencies) as part of every PR review.

Anti-Patterns This Skill Knows to Avoid

• ❌ Never commits directly to main/master branch — always works through PRs
• ❌ Never force-pushes to shared branches
• ❌ Never approves its own PRs without human review
• ❌ Never merges with failing CI — investigates failures first
• ❌ Never stores secrets, API keys, or credentials in code or commit messages
• ❌ Never ignores .gitignore patterns or commits generated files

Pairs Well With

• AI Cofounder ($49) — Engineering management within broader company operations
• Sub-Agent Orchestrator ($14) — Coordinate code generation across multiple repos or services
• Agent Memory Architecture ($1.99) — Remember codebase decisions and architectural context across sessions

Setup

Install in under 5 minutes. Drop the SKILL.md into your OpenClaw workspace and your agent gains full GitHub engineering capability immediately. Connect your GitHub token and point it at your repos.