AutoPR Slack AI — Serverless Autonomous AI Developer

AutoPR Slack AI is an enterprise-grade serverless autonomous AI developer pipeline. It accepts developer task requests directly from Slack or Telegram, resolves target repositories using Gemini intent classification and preflight context resolution, constructs an AST knowledge graph via graphify to optimize context efficiency, and executes the Google Antigravity Engine (agy) within GitHub Actions to generate, test, and open verified Pull Requests automatically.

---

---

Key Features

• High Speed & Deterministic Reliability: Combines sub-second Cloudflare Edge routing with preflight context resolution to deliver fast, reliable, end-to-end task completion.
• Graphify Knowledge Graph Integration: Automatically builds a persistent AST knowledge graph of the codebase prior to engine invocation. This pre-indexing provides targeted structural context, cutting LLM token usage by up to 50%.
• Serverless Edge Webhook Handling: Cloudflare Workers intercept incoming Slack slash commands and Telegram webhooks with sub-second response times.
• AI Intent & Repository Resolution: Integrates Gemini models to parse unstructured prompt text, determine target GitHub repositories, and resolve default branches.
• Headless Antigravity Engine Execution: Runs the Google Antigravity (agy) CLI inside ephemeral GitHub Actions runner VMs with session authentication.
• Automated Pull Request Lifecycle: Automatically creates target feature branches, applies precision code modifications, runs verification, and submits GitHub Pull Requests.
• Threaded Status Feedback: Updates Slack and Telegram discussion threads in real-time with execution status, PR links, and AI summarization.
• Interactive Sandbox & Web UI: Includes a local web interface (index.html, style.css, app.js) for simulating slash commands, inspecting architecture flows, and testing deployment configurations.

---

Specialized Agent Skills & Tooling

This project leverages specialized AI agent skills to achieve enterprise code quality, structural understanding, and design standards:

• Graphify (graphify): Analyzes codebase structure to build a persistent knowledge graph, reducing prompt token overhead by up to 50% and improving context resolution.
• UI/UX Pro Max (nextlevelbuilder/ui-ux-pro-max-skill): Provides design system token architecture, modern typography pairings, responsive layouts, and professional UI styling patterns.
• Agent Skills for Scalable Backend (addyosmani/agent-skills): Enforces clean architecture, decoupling service and repository layers for scalable Python backend development.

---

System Architecture

+---------------------+
| Slack / Telegram    |
| (Slash Command)     |
+----------+----------+
           | Webhook HTTP POST
           v
+---------------------+
| Cloudflare Worker   |  <-- Fast-path validation & Gemini intent parsing
+----------+----------+
           | Repository Dispatch API Event
           v
+---------------------+
| GitHub Actions VM   |
| (Runner Engine)     |
|                     |
| ├── 1. Environment & Auth Restoration (AGY_AUTH_CONFIG)
| ├── 2. Preflight Target Resolution (automation/preflight.py)
| ├── 3. Graphify Knowledge Graph Construction (Cuts token usage by 50%)
| ├── 4. Antigravity CLI Execution (agy run -y "$PROMPT")
| ├── 5. Automated Verification & Git Commit
| └── 6. Pull Request Submission (gh pr create)
+----------+----------+
           | Execution Result Callback
           v
+---------------------+
| Slack Thread Reply  |  <-- Posts PR URL & AI summary back to thread
+---------------------+

---

Project Structure

github_automation/
├── .github/
│   └── workflows/
│       ├── ai-autonomous-developer.yml   # Primary GitHub Actions execution workflow
│       ├── deploy-pages.yml             # GitHub Pages deployment workflow
│       └── test-workflow.yml            # Integration test workflow
├── automation/
│   ├── core/                            # Core engine configurations & logging
│   ├── domain/                          # Business entities and value objects
│   ├── interfaces/                      # API clients and GitHub/Slack adapters
│   ├── services/                        # Intent routing, preflight resolution, summarization
│   ├── preflight.py                     # Repository & target branch preflight script
│   └── main.py                          # Automation entrypoint
├── cloudflare-worker/
│   ├── worker.js                        # Worker entrypoint for dispatching events
│   ├── slack-worker.js                  # Slack webhook & challenge handler
│   └── wrangler.toml                    # Cloudflare Worker configuration manifest
├── product_demo/
│   ├── brag.mp4                         # Demonstration video
│   └── brag.jpg                         # Video poster frame
├── pyproject.toml                       # Python project configuration (uv / hatchling)
├── index.html                           # Sandbox Web UI structure
├── style.css                            # Sandbox Web UI styling
└── app.js                               # Sandbox Web UI interactive logic

---

Prerequisites

• Python: Version 3.12 or higher.
• Package Manager: uv (recommended) or standard pip.
• Node.js: Version 18 or higher (for Cloudflare Wrangler CLI).
• GitHub CLI: gh CLI installed and authenticated with repository permissions.
• Antigravity CLI: agy executable installed on the worker environment or GitHub Actions runner.

---

Configuration & Environment Secrets

1. GitHub Repository Secrets

Configure the following secrets under Settings > Secrets and variables > Actions in your GitHub repository:

Secret Name	Description
AGY_AUTH_CONFIG	Base64-encoded session configuration file (~/.gemini/config) for Google Antigravity CLI authentication.
PAT_TOKEN	GitHub Personal Access Token (Classic) with repo, workflow, and write:packages scopes.
SLACK_BOT_TOKEN	Slack Bot User OAuth Token (xoxb-...) for sending thread updates and status messages.
GEMINI_API_KEY	Optional API key for Gemini models used during preflight intent resolution.

2. Cloudflare Worker Secrets

Set worker secrets using wrangler secret put:

cd cloudflare-worker
npx wrangler secret put GITHUB_PAT
npx wrangler secret put SLACK_BOT_TOKEN
npx wrangler secret put GEMINI_API_KEY

---

Quickstart (Zero-Clone Setup via npx)

You can set up AutoPR in any repository with a single command — no manual cloning required:

npx autopr-slack

The interactive wizard will automatically:

1. Validate GitHub CLI & environment diagnostics.
2. Inject the reusable GitHub Actions workflow (.github/workflows/autopr.yml).
3. Deploy the Cloudflare Worker serverless edge webhook.
4. Generate your 1-click Slack App Manifest (slack-app-manifest.json).
5. Provision required repository secrets via gh secret set.

---

Manual Deployment & Setup

Deploying the Cloudflare Worker Manually

1. Navigate to the worker directory:

   cd cloudflare-worker
   

2. Install dependencies:

   npm install
   

3. Deploy to your Cloudflare account:

   npx wrangler deploy
   

4. Copy the output Worker HTTP endpoint URL and configure it as your Request URL in the Slack App settings under Slash Commands (e.g. /code or /autopr).

---

Local Development & Testing

Running Python Tests

Execute test suites using uv:

uv run pytest

Testing Preflight Target Resolution Locally

Run the preflight resolution module locally to verify repository intent parsing:

uv run python -m automation.preflight --prompt "bhaveshupadhyay/app Add Redis caching to user service"

Running the Cloudflare Worker Locally

Start local Wrangler environment:

cd cloudflare-worker
npx wrangler dev

---

Future Enhancements & Roadmap

The execution architecture is designed with decoupled provider interfaces to support multi-engine execution driver adapters and extended tool protocols in upcoming releases:

• Figma Model Context Protocol (MCP) Integration: Connecting Figma MCP servers to automatically generate production-grade UI code directly from Figma design frames and component specs.
• Multi-CLI Engine Integration: Extending beyond the Google Antigravity (agy) CLI to support Claude Code, OpenCode, and other autonomous developer CLI tools.
• Pluggable Engine Selection: Allowing users to specify execution engines via slash parameters (e.g., /code --engine=claude or /code --engine=opencode).
• Distributed Knowledge Graph Caching: Persisting and reusing graphify knowledge graph structures across workflow runs to further optimize cold-start execution speeds.

---

Contributing Guidelines

We welcome contributions to AutoPR Slack AI. Please follow these guidelines when submitting pull requests or opening issues:

Branching Strategy

• main: Production-ready branch. All changes enter via Pull Requests.
• Feature branches: Use prefix feat/ (e.g., feat/add-telegram-adapter).
• Bug fix branches: Use prefix fix/ (e.g., fix/intent-router-fallback).
• Documentation: Use prefix docs/ (e.g., docs/update-architecture-spec).
• Refactoring: Use prefix refactor/ (e.g., refactor/clean-architecture).

Pull Request Process

1. Fork the repository and create a new feature branch from main.
2. Ensure code follows clean architecture patterns, PEP 8 standards, and includes proper type hints.
3. Write or update unit tests for any new or modified functionality.
4. Run uv run pytest to ensure all tests pass.
5. Submit a Pull Request detailing the problem solved, changes made, and verification steps taken.
6. Obtain approval from at least one repository maintainer before merging.

---

License & Security

This project is distributed under the MIT License. For security vulnerabilities or concerns, please open an issue or contact the maintainers directly.