Stirrup is a lightweight Python framework for building AI agents, published by Artificial Analysis. It is designed as either a package or a starting template. Its approach is to let the model choose how to complete a task, similar to Claude Code, rather than imposing a rigid workflow. The authors say they studied leading agents to build in practices for context management and foundational tools.
Built in capabilities include online search and web browsing, code execution locally, in a Docker container or in an E2B sandbox, an MCP client, and document input and output. A skills system lets developers add modular, domain specific instruction packages, and a generic Tool class makes custom tools easy to define. Context management automatically summarizes conversation history as it nears the limit. Pre built clients cover OpenAI compatible APIs and LiteLLM, and developers can bring their own client. Multimodal support processes images, video and audio with automatic format conversion.
Stirrup is installed with pip or uv, with optional extras for LiteLLM, Docker, E2B, MCP and browser tools. It is aimed at developers who want a minimal base for custom agents. The documentation site publishes no pricing, and web search in the default tools requires a separate Brave API key.
Related Tools
Other development & code tools in the same category
