aden-hive/hive

Health v2 · maintenance only

Adoption not part of the score

10974 stars · 5672 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

OpenHive is a model-agnostic multi-agent harness and runtime layer for production AI workloads, organizing specialized agents into 'colonies' led by a persistent Queen agent. It provides state management, failure recovery, observability, and human-in-the-loop oversight, with MCP tool support and support for OpenAI, Anthropic, and Gemini models.

Use cases

• orchestrate multiple ai agents to run a business process
• run autonomous agents in production with failure recovery
• add human approval steps to agent workflows
• build self-improving agent systems with mcp tools
• coordinate worker agents under a lead agent
• self-host an agent runtime for claude and openai models

When to choose

• you need production-grade agent orchestration with state management and observability
• you want multi-agent collaboration with human-in-the-loop oversight
• you need a model-agnostic runtime supporting OpenAI, Anthropic, and Gemini
• you want MCP tool integration in an agent framework

When to avoid

• you only need a simple single-agent chatbot
• you want a fully managed cloud service rather than self-hosting
• your use case is unrelated to LLM-driven agents

Facets

framework · maturity active

agent-framework workflow-automation mcp chatbot monitoring artificial-intelligence large-language-models developer-tools python self-hosted cross-platform multi-agent human-in-the-loop self-improving model-agnostic agent-colony harness production-ai openai anthropic gemini ai-agents automation

Member repositories

For agents

markdown · JSON · MCP: product_card(name="aden-hive/hive")