Aury Agents

A flexible Agent framework supporting both React Agent (autonomous loop) and Workflow (DAG orchestration) patterns.

Features

• React Agent: Autonomous agent with tool calling loop
• Workflow: DAG-based orchestration for complex pipelines
• Session Management: Event-sourced session with snapshot and revert
• Permission System: Fine-grained tool execution control
• Memory System: Semantic memory with vector storage
• Compaction: Token-efficient session history management

Installation

pip install aury-agents

Quick Start

from aury.agents import ReactAgent, Tool

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

# Create agent
agent = ReactAgent(
    tools=[search],
    llm_provider=your_llm_provider,
)

# Run
result = await agent.run("What is the weather today?")

Development

# Install dev dependencies
uv sync

# Run tests
pytest

# Type check
mypy aury

# Lint
ruff check aury

License

MIT