hive

What it is

Hive is a multi-agent harness for production workloads that dynamically generates multi-agent topologies to tackle complex, long-running business workflows. It emphasizes state management, fault tolerance, observability, and human oversight.

How it works

• Define a goal in plain English.
• The agent graph is auto-generated, including connection code and test cases.
• Workers execute using an SDK-wrapped node structure with observability and tool access.
• A control plane provides real-time metrics and policy management.
• If failures occur, the graph can evolve and redeploy.

Getting started

Installation guidance emphasizes a quickstart workflow rather than a standard pip install. The repository recommends cloning and running a quickstart script:

# Clone the repository
git clone https://github.com/aden-hive/hive.git
cd hive

# Run quickstart setup (macOS/Linux)
./quickstart.sh

# Windows (PowerShell)
.\quickstart.ps1

This setup creates framework and tools virtual environments, a credential store, and an LLM configuration, then opens the Hive interface in a browser. Tip: to reopen the dashboard later, run hive open from the project directory.

Build Your First Agent

Type the agent you want to build in the home input box; the queen assists in building the solution.

Use Template Agents

Choose a sample agent template to run or extend.

Run Agents

Select an agent and use the Run button or interact with the queen to execute the agent.

Integration

Hive is model-agnostic and supports various providers via LiteLLM, including Anthropic, OpenAI, OpenRouter, Hive LLM, and local models. It also connects to business systems (CRM, support, data, APIs) via MCP.

Documentation

Provides links to developer guides, getting started, and configuration.

Licensing and contribution

Open-source under the Apache License 2.0; contribution guidelines and issue-based workflow are described.