auggie-mcp
Run Augment Code as a coding agent via the Auggie CLI
Documentation
Auggie MCP Server
Minimal MCP server exposing Auggie CLI as tools for Q&A and code implementation.
Tools
- ask_question: Repository Q&A via Auggie’s context engine.
- implement: Implement a change in the repo; dry-run by default.
Requirements
- Node.js 18+
- Python 3.10+ available on the system (used internally; no manual setup needed)
- Auggie CLI installed (check by running `auggie --version`) - _see installation guide_
Authentication (AUGMENT_API_TOKEN)
Retrieve your token via the Auggie CLI:
# Ensure Auggie CLI is installed and on PATH
auggie --version
# Sign in (opens browser flow)
auggie login
# Print your token
auggie --print-augment-tokenProvide the token in either of these ways:
- Cursor/Claude config (recommended): set it under `env` for the server
{
"mcpServers": {
"auggie-mcp": {
"command": "npx",
"args": ["-y", "auggie-mcp@latest"],
"env": { "AUGMENT_API_TOKEN": "YOUR_TOKEN" }
}
}
}- Shell environment (macOS/Linux)
One-off for a single command:
AUGMENT_API_TOKEN=YOUR_TOKEN npx -y auggie-mcp --setup-onlyPersist for future shells (zsh):
echo 'export AUGMENT_API_TOKEN=YOUR_TOKEN' >> ~/.zshrc
source ~/.zshrcSecurity tip: never commit tokens to source control. Prefer per-machine environment variables or your client's secure config store.
Configure Clients
Cursor via npx
Use this MCP config in Cursor (global or per-project):
{
"mcpServers": {
"auggie-mcp": {
"command": "npx",
"args": ["-y", "auggie-mcp@latest"],
"env": { "AUGMENT_API_TOKEN": "YOUR_TOKEN" }
}
}
}This will:
- download the wrapper package,
- create a local Python venv inside the package,
- install `requirements.txt`, and
- launch the Python server in `stdio` mode.
Quick test via npx (terminal)
# Install deps into the package's local venv (no global installs)
npx -y auggie-mcp --setup-only
# Run the server (stdio). Useful for quick smoke-tests.
npx -y auggie-mcp
# Optional: start HTTP mode for manual debugging
npx -y auggie-mcp -- --httpClaude Desktop (macOS)
Edit `~/Library/Application Support/Claude/claude_desktop_config.json` and add:
{
"mcpServers": {
"auggie-mcp": {
"command": "npx",
"args": ["-y", "auggie-mcp@latest"],
"env": { "AUGMENT_API_TOKEN": "YOUR_TOKEN" }
}
}
}Security and permissions
- Default: `implement` runs in dry‑run mode. No files are written, no shell runs; you get a proposed diff.
- Enable writes: set `dry_run: false`.
- Recommendation: use a feature branch and review the diff before merging.
Frequently asked questions
What is auggie-mcp?
auggie-mcp is Run Augment Code as a coding agent via the Auggie CLI
How do I install auggie-mcp?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is auggie-mcp open source?
Yes — it is hosted on GitHub at https://github.com/saharmor/auggie-mcp and has 1 stars.
Related MCP tools
基于大模型搭建的聊天机器人,同时支持 微信公众号、企业微信应用、飞书、钉钉 等接入,可选择ChatGPT/Claude/DeepSeek/文心一言/讯飞星火/通义千问/ Gemini/GLM-4/Kimi/LinkAI,能处理文本、语音和图片,访问操作系统和互联网,支持基于自有知识库进行定制企业智能客服。
An LLM agent that conducts deep research (local and web) on any given topic and generates a long report with citations. Built for the Model Context Protocol to
🚀 The fast, Pythonic way to build MCP servers and clients Trusted by 19900+ developers. Trusted by 19900+ developers. Trusted by 19900+ developers.
🔥 MaxKB is an open-source platform for building enterprise-grade agents. MaxKB 是强大易用的开源企业级智能体平台。 for the Model Context Protocol. Enhance AI assistants with po
Expose your FastAPI endpoints as Model Context Protocol (MCP) tools, with Auth! Python-based implementation. Trusted by 11000+ developers.
Agent Framework For Fintech for the Model Context Protocol. Enhance AI assistants with powerful integrations. Python-based implementation.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP