The Autonomous Deployment Pipeline
Atlas analyzes your project, runs the build, uses an LLM to auto-fix build errors, and deploys to a cloud provider—all in a single command.
- 🧠 Auto-Healing Builds: When a build fails, Atlas diagnoses the error, suggests a patch using an LLM (Anthropic, Mistral, Groq, etc.), and retries autonomously.
- 🚀 One-Command Deploy: From local directory to live URL without opening the browser.
- 🔌 Multi-Provider Support: First-class deployments for Vercel, Render, and Netlify.
- 🔎 Framework Detection: Automatically detects React, Next.js, Express, and Vite projects and runs the correct build tools.
- 🛡️ Secure Credential Management: Employs the OS keychain (via
go-keyring) to store your sensitive LLM and Provider API keys safely. - 🤖 Deterministic Pipeline: Not a wild looping agent. Atlas follows a strictly bounded, predictable state machine for safe deployments.
You can install Atlas using our seamless installation scripts:
macOS & Linux:
curl -sSL https://raw.githubusercontent.com/Yashh56/atlas/master/install.sh | shWindows (PowerShell):
iwr -useb https://raw.githubusercontent.com/Yashh56/atlas/master/install.ps1 | iexOr via Go directly:
go install github.com/Yashh56/atlas/cmd/atlas@latestAtlas requires an LLM to fix broken builds. We support Anthropic, OpenAI, Mistral, Gemini, Groq, and xAI.
# Example: Store your Anthropic key securely via the CLI wizard
atlas models set anthropicYou can also use environment variables like export ANTHROPIC_API_KEY=...
Navigate to any supported project directory and let Atlas do the magic:
cd ./my-nextjs-app
atlas .Atlas will interactively ask you to choose a provider (e.g., Vercel), authenticate if you aren't already, run the build, fix any errors, and give you a live URL.
(For CI/CD usage, you can run non-interactively: atlas . --action deploy --provider vercel)
| Command | Description |
|---|---|
atlas <path> |
Interactive deployment wizard |
atlas <path> --action deploy --provider <name> |
Full non-interactive pipeline |
atlas providers |
Check deploy provider auth status |
atlas models |
Check LLM API key status |
atlas testllm <path> |
Verify LLM key with a live ping |
For a full breakdown of the architecture, see our Documentation.
| Framework | Status |
|---|---|
| NextJS | ✅ |
| React | ✅ |
| Vite | ✅ |
| Express | ✅ |
| Django | ⏳ Planned |
| Provider | Status |
|---|---|
| Vercel | ✅ |
| Render | ✅ |
| Netlify | ✅ |
| Fly.io | ⏳ Planned |
| Railway | ⏳ Planned |
Q: Does Atlas rewrite my code unexpectedly?
A: No. Atlas uses exact-match patching. It does not rewrite entire files and will explicitly ask for approval before continuing the deployment after a patch.
Q: Where are my API keys stored?
A: Atlas stores credentials securely in your operating system's native keychain (Keychain on macOS, Credential Manager on Windows, Secret Service on Linux) and never writes them to plaintext files.
Q: Can I use local models?
A: Yes! You can configure Atlas to use local LLMs (like Ollama) for completely free, private auto-healing.
- Docker Support: Automatically generate
Dockerfiles for unknown frameworks. - Rollback Mechanics: Automatically revert to a previous healthy deployment if the post-deployment health check fails.
- More Frameworks: Deep integrations for Django, Spring Boot, and Laravel.
- Multi-Environment Deployments: Native support for staging and production branching.