autoweave 0.1.2

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

AutoWeave: Multi-Agent Orchestration Library

AutoWeave is the robust 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.

Core Features

• Workflow Orchestration: Define, compile, and execute Directed Acyclic Graphs (DAGs) of agentic tasks.
• Durable State: Resume paused runs, track individual attempts, and safely persist context to PostgreSQL.
• Human-in-the-Loop: Built-in primitives to pause execution and request approvals or clarifications from a human.
• Queue Dispatch: Offload long-running autonomous tasks to Celery workers backed by Redis.
• Local Monitoring: Inspect active and historic runs via a beautiful, lightweight local developer dashboard.

Installation

Install AutoWeave directly from PyPI using pip or uv:

pip install autoweave

Quick Start (CLI)

AutoWeave provides a comprehensive CLI for local orchestration.

1. Initialize a new project in a fresh directory:

autoweave new-project

2. Bootstrap the local environment and configuration files:

autoweave bootstrap

3. Start the control plane UI and background Celery worker:

autoweave start

4. Execute a workflow from the terminal:

autoweave run-workflow --root . --request "Create a Python script that calculates Fibonacci numbers"

Programmatic Usage

AutoWeave exposes a clean Python API for integrating orchestration into your own applications.

1. Launching a Workflow

from autoweave.orchestration.runtime import build_local_runtime

# Initialize the runtime for your project directory
runtime = build_local_runtime(root_path="./my-weave-project")

# Dispatch a new workflow to the agents
workflow_run = runtime.launch_workflow(
    request="Review the backend contract and propose next steps"
)
print(f"Successfully started workflow run: {workflow_run.id}")

2. Inspecting the State

from autoweave.monitoring.service import MonitoringService

service = MonitoringService(db_path="./my-weave-project/autoweave.db")

# Fetch a snapshot of all active runs
state_snapshot = service.snapshot(limit=5)
print(state_snapshot)

Dashboard & Monitoring

AutoWeave includes a built-in monitoring dashboard to trace agent executions, view generated artifacts, and resolve human-in-the-loop approvals.

To start it independently:

autoweave ui --root ./my-weave-project

Then navigate to http://localhost:8766 in your browser.

Support & Architecture

For comprehensive architecture specs, deployment instructions, and advanced configuration options, visit our GitHub Repository Documentation.