H

Hive

Multi-Agent Harness for Production AI

open-sourceagent-frameworks
11.1k
Stars
+925
Stars/month
22
Commits (90d)
8
Releases (6m)

Star Growth

+2.8k (33.3%)estimated from velocity
8.2k9.7k11.3kJul 2Sep 30

Overview

OpenHive is a zero-setup, model-agnostic runtime for colonies of agents where a Queen agent spawns worker clones for parallel execution. It provides crash-safe state management, human oversight, cost enforcement, and deep observability built into a single execution primitive. The system coordinates agents through a shared ledger and persistent plan without requiring graph compilation or orchestration boilerplate.

Deep Analysis

Key Differentiator

Uses a single execution primitive where the Queen agent is an agent loop and every worker is a clone of it, coordinating through shared ledger without graph compilation.

⚡ Capabilities

  • • multi-agent colonies
  • • parallel execution
  • • state management
  • • failure recovery
  • • human-in-the-loop oversight
  • • cost enforcement
  • • observability
  • • crash-safe park/resume

🔗 Integrations

custom model supportgeneral compute usebrowser use with native extension

✓ Best For

  • ✓ teams moving AI agents from prototype to production
  • ✓ executing real business processes with AI agents
  • ✓ production workloads requiring uptime and auditability
  • ✓ adaptive agents that improve over time through reflexion and memory

✗ Not Ideal For

  • ✗ simple agent chains or one-off scripts
  • ✗ experimenting with basic agent demos

⚠ Known Limitations

  • ⚠ README mentions it may not be best fit for simple agent chains or one-off scripts

Alternatives

See all 8 Hive alternatives →

Compare Hive

Maintain Hive?

Show your live rank in your README, or put Hive in front of every visitor to AgentoolRank.