Fast Agent
An open-source terminal coding agent and Python toolkit that supports local models through llama.cpp, cloud providers, and MCP tools.
Alternative to:
Fast Agent is a terminal coding assistant and Python toolkit for developers building AI agents and automated workflows. You can choose local LLMs or hosted providers and use the same toolkit for interactive work, scripting, and evaluation.
- Model choice: Connect local models through the generic provider or llama.cpp servers, or use Anthropic, Google, and OpenAI-compatible endpoints.
- Tools and client connections: Give agents shell tools, MCP servers, or Python functions, and expose agents to other clients through ACP or MCP.
- Agent workflows: Run agents in sequence or in parallel, route requests to a suitable agent, or let an orchestrator divide tasks and combine results.
- Output evaluation: Pair a generator with an evaluator that provides feedback and requests revisions; agents can also ask a person for context during a task.
- Context and extensions: Use system prompt templates, Agent Skills, and prompt files, then extend behavior with Python plugins and hooks or share configurations through Card Packs.
The toolkit runs on your machine. Local model support includes automatic connection configuration for llama.cpp servers. Choosing a cloud provider or remote endpoint sends model requests to that service; connecting external tools may also involve remote services.
Multimodal prompts and MCP resources can carry content beyond text, with supported content types depending on the selected model provider. Fast Agent is open source under Apache 2.0.
