Polycode - AI-Powered GitHub Issue Automation

Overview

Polycode is a multi-agent automation system that integrates with GitHub to automatically process issues and code tasks using specialized AI crews.

How It Works

1. Issue Creation & Tagging

• Create issues in GitHub repositories as usual
• Apply specific labels to trigger different workflows
• Examples: ralph, bug-fix, feature-dev, security-audit

2. Automated Workflows

The system includes three main AI crews that automatically activate based on labels:

Bug Fix Crew

• Triages bug reports and reproduces issues
• Investigates root causes and implements fixes
• Creates pull requests with test coverage

Feature Development Crew

• Breaks down features into ordered user stories
• Implements features with automated testing
• Handles code reviews and documentation

Security Audit Crew

• Scans code for vulnerabilities
• Implements security fixes with regression tests
• Creates comprehensive security PRs

3. Label-to-Flow Mapping

• Configure which labels trigger which workflows
• Example: ralph label → ralph_flow workflow
• Supports repository-specific mapping patterns
• Prioritized workflow execution

4. GitHub App Integration

• Install Polycode as a GitHub App on repositories
• Webhook-driven automation (no manual polling)
• Secure authentication using GitHub App tokens
• Push capabilities to external repositories (if app is installed)

5. Process Flow
   GitHub Issue + Label → Webhook → Workflow Trigger → AI Crew Processing → Code Changes → Pull Request

6. Key Features

• Multi-repository support: Manage issues across multiple repos
• Automated execution: Workflows start instantly when labels are applied
• Code generation: AI agents create and push code automatically
• Testing integration: Includes automated regression and integration tests
• External repo access: Push changes to repositories you don't own (if app installed)

Getting Started

1. Install the GitHub App on target repositories
2. Configure label-to-flow mappings
3. Create issues and apply labels to trigger workflows
4. Monitor progress through webhook logs and the GitHub App interface

The system essentially transforms GitHub issue management into an automated development pipeline, where AI agents handle coding tasks while developers focus on higher-level planning and review.