fast-agent
A Python framework for building, running and evaluating LLM agents with full support for the Model Context Protocol, including sampling and elicitation, plus an interactive terminal and multiple model providers.
fast-agent is a framework for interacting with LLMs, usable as a coding agent, a development toolkit, or a platform for evaluation and workflow automation. It can be started as an interactive session with shell support, or configured with prebuilt packs for providers such as Hugging Face Inference or an OpenAI Codex plan.
It provides an optional terminal-style interactive prompt with completions and in-terminal menus, and streamed responses, alongside a declarative syntax for composing prompts and MCP servers into agents and workflows. It is described as the first framework with complete, end-to-end tested support for MCP features including Sampling and Elicitations, and can also be driven through any ACP-compatible client. Model support includes Anthropic, OpenAI, Google, Azure, Ollama and local models such as llama.cpp.
It installs with uv and is licensed under Apache-2.0.