AutoHeal is an industry-grade, autonomous CI/CD intelligence engine designed to detect, analyze, and repair pipeline failures before code even leaves the developer's machine. By providing a "pre-push" intelligence layer, it ensures that only healthy code enters the integration stage.
Modern CI/CD pipelines are often "black boxes" that fail for predictable reasons—dependency drifts, configuration errors, or subtle code conflicts. Developers shouldn't have to wait 20 minutes for a pipeline to fail just to see a common error log.
AutoHeal acts as an Autonomous Healing Layer. Currently, the platform allows developers to perform deep scans of repository structures and dependencies. It predicts potential failures in the Working Phase—identifying issues before a push is even initiated.
AutoHeal isn't just an API wrapper; it's built on a specialized Agentic Framework:
- Diagnostic Agent: Recursively scans thousands of files in milliseconds using the GitHub Git Trees API to build a complete mental model of the repository.
- Intelligence Engine: Leverages Gemini 2.5 Flash with highly specialized DevOps prompts to identify anti-patterns in
.github/workflows,package.json, and environment configs. - Healing Service: Automatically generates code-level patches for identified risks, providing developers with ready-to-apply fixes.
Analyze your repository context and CI configurations (.github/workflows, Jenkinsfile, etc.) to calculate a failure probability score before you initiate a deployment.
Utilizes high-performance recursive tree mapping to analyze entire repository structures in seconds, ensuring complete coverage of even the largest monorepos.
A comprehensive dashboard provides a unified view of your build history, success metrics, and average durations, allowing project leads to monitor health levels at a glance.
Fetch and analyze live GitHub Actions logs. The system identifies specific job failures and provides autonomous repair suggestions to keep the development cycle fluid.
Maintain a connected set of repositories and switch between them instantly to monitor the health of your entire portfolio from a single interface.
| Platform Layer | Technology |
|---|---|
| Intelligence Engine | High-performance Python backend with Pydantic-Settings |
| Agent Logic | Custom Autonomous Diagnostic Agent |
| Logic Framework | FastAPI (Asynchronous stream handling) |
| Interface | Next.js 16 with "Studio Zen" design system (Fluid animations) |
| AI Model | Google Generative AI (Gemini 2.5 Flash) |
- Node.js 22+ (matches our CI/CD environment)
- Python 3.13+
- GitHub Personal Access Token
- Google AI API Key
cd autoheal-backend
python -m venv venv
source venv/bin/activate # Windows: .\venv\Scripts\activate
pip install -r requirements.txtCreate a .env file in autoheal-backend/:
GITHUB_ACCESS_TOKEN=your_token
GEMINI_API_KEY=your_key
JWT_SECRET_KEY=your_secretRun the server: python main.py
cd autoheal-ci
npm install
npm run devAutoHeal is an open-source project and we welcome contributions from the community! Whether you are fixing a bug, suggesting a new feature, or improving the healing logic:
- Feel free to contribute!
- Fork the repository, create your feature branch, and submit a PR.
- Let's build the future of autonomous DevOps together.
Built for the next generation of DevOps Engineers.