autoweave 0.1.1

Terminal-first multi-agent orchestration library built around OpenHands remote workers and Vertex AI.

AutoWeave: Multi-Agent Orchestration Library

AutoWeave is the execution engine for multi-agent software engineering teams. It orchestrates specialized AI agents as a coherent team, managing workflow compilation, task graphs, queue-backed durable execution, and human-in-the-loop approvals.

What is AutoWeave?

AutoWeave owns the orchestration layer, explicitly decoupled from the single-agent execution layer (e.g., OpenHands).

• Workflow Orchestration: Define, compile, and execute DAGs of agentic tasks with dependency-aware dynamic scheduling.
• Durable State: Resume paused runs, track attempts, and persist context safely across PostgreSQL.
• Human-in-the-Loop: Native primitives for pausing execution to request approvals or clarifications.
• Queue Dispatch: Offload long-running tasks to Celery workers backed by Redis.
• Local Monitoring: Inspect runs via a lightweight local dashboard and playground.

Installation

You can install AutoWeave via pip:

pip install autoweave

Quick Start

Initialize a new AutoWeave project in a fresh directory:

autoweave new-project

Next, bootstrap the local environment and configurations:

autoweave bootstrap

Start the control plane UI and Celery worker:

autoweave start

CLI Reference

The autoweave CLI provides complete control over your local environment. Here are the core commands:

• status: Show a minimal repository status summary.
• validate: Validate docs, configs, and sample agent fixtures.
• create-agent: Create a new agent bundle with soul, playbook, config, and skills.
• run-workflow: Run the current workflow from a user request.

Run autoweave --help for the full list of options.

Support

For comprehensive architecture specs and deployment instructions, visit our GitHub repository documentation.