DevPilot AI

DevPilot AI is a CLI-first local developer environment setup assistant.

It clones unfamiliar repositories, detects the stack deterministically, installs dependencies, starts services in managed terminals, captures logs, and uses AI only for explanations, repository chat, and debugging help.

Features

• devpilot setup <github-url> for end-to-end repository onboarding
• Deterministic detection for languages, frameworks, package managers, Docker, CI/CD, databases, and monorepos
• Managed service runtime for frontend, backend, database, workers, and queues
• Local-first configuration under ~/.devpilot/
• Provider-independent AI layer with Gemini, OpenAI, and OpenRouter support
• Repository intelligence with chunking, embeddings, ChromaDB retrieval, and grounded chat
• Local security controls for command risk, workspace isolation, path protection, and audit logs

Installation

pip install devpilot-labs

Recommended full install:

pip install "devpilot-labs[all]"

With uv:

uv tool install "devpilot-labs[all]"

For local development:

python -m pip install -e ".[dev]"

Quick Start

devpilot auth login
devpilot setup https://github.com/aarjjun/LinkedInSpeak
devpilot ps
devpilot logs frontend
devpilot chat

CLI Commands

Configuration

DevPilot stores local user data outside the installed package:

~/.devpilot/
  config.json
  credentials.enc
  workspace/
  logs/
  history/
  database/
  cache/
  indexes/

API keys are never written into repositories.

Supported Providers

• Gemini: pip install "devpilot-labs[gemini]"
• OpenAI: pip install "devpilot-labs[openai]"
• OpenRouter: pip install "devpilot-labs[openrouter]"
• Everything: pip install "devpilot-labs[all]"

Supported Frameworks

DevPilot uses deterministic strategies for common Python, Node.js, Docker, and compiled-language projects, including FastAPI, Flask, Django, React, Vite, Next.js, Express, Spring Boot, Go, Rust, .NET, Docker Compose, and workspace-style monorepos.

Architecture

CLI
  -> Workflows
  -> Agents
  -> Tools
  -> Operating System

The CLI is the product. Any local API/dashboard code is optional visualization and must not contain setup business logic.

Development

python -m pip install -e ".[dev]"
python -m ruff check .
python -m mypy
python -m pytest
python -m build

Troubleshooting

Run:

devpilot self-test
devpilot doctor
devpilot logs

If AI commands fail, check:

devpilot auth status
devpilot auth switch

Contributing

See CONTRIBUTING.md.

License

MIT. See LICENSE.