pureagents

LLM agents without the complexity. Powerful, but simple.

Philosophy

Most agent frameworks are over-engineered. Thousands of lines, dozens of abstractions, patterns on patterns.

pureagents is different:

• Powerful: All the features you need, from basics to production
• Simple: ~2,200 lines. Read the entire codebase in 30 minutes
• Elegant: Clean API. One decorator, one class. No magic
• Buildable: Fork it, modify it, build on top. Not a black box

Many features ≠ complicated. Every feature is optional and modular.

Features

Core

Feature	Usage
Tools	@tool decorator
Providers	provider="openai"
Streaming	agent.stream()
Memory	session="my-chat"
Structured outputs	output=MyDataclass
Hooks	on_tool_call=fn
Templates	template="coder"
Batch	agent.batch([...])

Reliability

Feature	Usage
Retry	retries=3
Timeout	timeout=30.0
Context limit	max_messages=20
Fallback	fallback="mistral"
Caching	cache=True

Tools

Feature	Usage
Tool timeout	@tool(timeout=10)
Tool groups	@tool(group="web")
Force tool use	tool_choice="required"
Parallel execution	Automatic

Orchestration

Feature	Usage
Chaining	chain([agent1, agent2], prompt)
Routing	Router(agents={...})
Planning	plan=True
Graph	Graph() with conditional edges

Advanced

Feature	Usage
Images	images=["photo.jpg"]
Validation	validator=fn
Usage tracking	agent.usage

Quick example

from pure_agents import Agent, tool


@tool
def search(query: str) -> str:
    """Search the web."""
    return f"Results for {query}..."


agent = Agent(tools=[search])
result = await agent.run("Find the weather in Madrid")

Installation

pip install pureagents

export MISTRAL_API_KEY="your-key"  # Default provider