selectools
selectools is a Python framework for building and orchestrating AI agents. It solves the complexity of multi-agent workflows by providing a simple, code-first approach with no DSL or compile step. It is for developers who want to build agentic applications using familiar Python patterns and support for multiple LLM providers.
✨ Key features
- Build single agents or multi-agent graphs with plain Python.
- Compose pipelines using the pipe operator (|).
- Deploy agents with a single command: selectools serve agent.yaml.
- Supports OpenAI, Anthropic, Gemini, and Ollama providers.
- Stable API with @stable symbols and compatibility promise.
- Includes guardrails, structured output, and streaming contracts.
🎯 Use cases
- Create a single agent with tools for calculations or web search.
- Orchestrate a chain of agents for tasks like writing and reviewing a blog post.
- Deploy an agent as a service using a YAML configuration.
- Build conversational agents with structured output and streaming.
- Implement scheduled agents that run on cron or interval schedules.
🚀 Usage
from selectools import Agent, OpenAIProvider
agent = Agent(tools=[search, calculate], provider=OpenAIProvider())
result = agent.run("What is 15 * 7?")
❓ FAQ
What Python versions are supported?
Python 3.10+ is required; Python 3.9 is no longer supported.
Which LLM providers are supported?
OpenAI, Anthropic, Gemini, and Ollama are supported.
Is the API stable?
Yes, selectools is 1.0 and the public API is frozen; @stable symbols have a compatibility promise.
How do I deploy an agent?
You can deploy with the command: selectools serve agent.yaml.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.