Fast Agent

What is Fast Agent?

Fast Agent is a Python framework for building and interacting with sophisticated AI agents and workflows in minutes. It runs on any system with Python and the uv package manager, and supports both Anthropic (Haiku, Sonnet, Opus) and OpenAI models (gpt‑4o/gpt‑4.1, o1/o3). It is the first framework with complete, end‑to‑end tested MCP feature support, including Sampling.

How to use Fast Agent?

Install via uv pip install fast‑agent‑mcp, then run uv run fast‑agent setup to create example agent and config files. Define agents in Python files using the @fast.agent decorator and execute them with uv run agent.py. Switch models with --model, target specific agents with --agent, send a one‑shot message with --message, and suppress progress output with --quiet.

Key features of Fast Agent

• Declarative syntax for composing agents and workflows
• Complete MCP support including Sampling and Roots
• Multi‑modal: supports images and PDFs from both Anthropic and OpenAI
• Built‑in workflow patterns: chain, parallel, evaluator‑optimizer, router, orchestrator
• Agents can request human input during task execution
• Model selection at runtime for testing interoperability
• Passthrough and playback LLMs for rapid development and testing

Use cases of Fast Agent

• Build a social media post writer by chaining a URL summarizer and a post generator
• Create a researcher agent that uses an evaluator‑optimizer loop to refine reports
• Develop a data analysis agent that acts like a ChatGPT‑style assistant with MCP Roots
• Parallelize translations into multiple languages using fan‑out/fan‑in workflows
• Prototype and test Python glue code for AI applications with playback LLMs

FAQ from Fast Agent

Which models does Fast Agent support?

Fast Agent supports Anthropic models (Haiku, Sonnet, Opus) and OpenAI models (gpt‑4o/gpt‑4.1 family, o1/o3 family). Models can be specified per‑agent or switched at runtime with the --model flag.

How do I install Fast Agent?

Install with uv pip install fast‑agent‑mcp. You need the uv package manager for Python. After installation, run uv run fast‑agent setup to create example files.

Does Fast Agent support multimodal input?

Yes. Fast Agent handles images and PDFs via prompts, resources, and MCP tool call results for both Anthropic and OpenAI endpoints.

Can agents request human input?

Yes. Setting human_input=True on an agent allows it to prompt the user for additional context or assistance during task completion.

Is Fast Agent open source?

Yes, Fast Agent is licensed under the MIT License (see the LICENSE file in the repository).