fast-agent CLI
A terminal agent and Python framework with MCP, SKILL.md and ACP support for coding, building and evaluating your own agents
- 3,951 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- In shell mode the agent runs arbitrary commands
- Sends data to the chosen model provider
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/fast-agentDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- In shell mode the agent runs arbitrary commands
- Sends data to the chosen model provider
This is third-party code. Review the repository files before installing.
What it does
fast-agent runs in the terminal as a coding agent with shell access and built-in subagents. The /skills command adds skills from its registry, and /connect attaches MCP servers over stdio or streamable HTTP with OAuth. As a framework it lets you declare agents and workflows in Python and test them. It supports Anthropic, OpenAI and Google models, local models via llama.cpp and Ollama, and any setup can be served to an ACP client.
Who it is for. For developers who build their own MCP-based agents or want an open terminal agent with a free choice of model.
Good fit when
- You need to build and test an MCP agent or workflow quickly
- You want a terminal agent that works with many providers and local models
- You need to debug an MCP server, including streamable HTTP transport and elicitations
Not a fit when
- You are happy with Claude Code or Codex and do not write your own agents
- You need a ready-made skill rather than an agent runtime
Example request
Connect the huggingface MCP server and find speech recognition models for RussianLimitations
Requires uv and Python. Cloud models need provider keys; the OpenAI and Anthropic APIs are not reachable from Russia without a VPN, while local models work offline. Shell mode runs commands on your machine.
How to disable. Remove the tool with uv tool uninstall fast-agent-mcp.
FAQ
Can I try it without installing?
Yes, uvx fast-agent-mcp@latest -x starts an interactive shell session without installation.
Does it work with local models?
Yes, there is a generic provider for local models, and fast-agent model llamacpp writes the llama.cpp config for you.
Discussion
No questions or reviews yet.
Related
CLI toolsA Python framework and CLI for MCP: write servers as plain functions, test tools and install servers into Claude Code, Cursor and other clients
MCP Agent Mail
mcp_agent_mail
A mail coordination layer for coding agents: agents have memorable names, inboxes and outboxes, history search and voluntary file reservations
A Python framework and CLI for building MCP servers and tools with built-in OAuth authorization
A universal MCP CLI client from Apify: persistent sessions, OAuth, JSON output for scripts and shell access to any MCP server for agents
A zero-dependency Python CLI: one catalog of MCP servers and skills for every agent, with only a compact index in context
A CLI that turns any MCP server, OpenAPI spec or GraphQL endpoint into commands at runtime, with no codegen
README in short
The README suggests starting with uvx fast-agent-mcp@latest -x or installing via uv tool install. It covers packs for Hugging Face Inference Providers and the Codex plan, attaching MCP servers and managing skills. It highlights full MCP feature support, including sampling, elicitations and transport diagnostics. Docs live at fast-agent.ai and in the docs folder. Apache-2.0 licensed.