AI BriefingOpenHarness - An Open Agent Harness with Built-in Personal Agent Ohmo
Key point
OpenHarness, a lightweight open-source Python infrastructure that turns LLMs into real agents, has been released.
Details
A lightweight open-source Python infrastructure that turns LLMs into real agents with integrated tool use, skills, memory, and multi-agent coordination.
It provides 43 built-in tools (File I/O, Shell, Search, Web, MCP, etc.), supporting Pydantic validation, parallel execution, permission checks, and Hook functionality.
The built-in personal agent ohmo converses on Slack, Telegram, Discord, and more, automatically writing code, running tests, and opening PRs. It leverages existing Claude Code or Codex subscriptions, so no separate API key is needed.
Key features:
- Compatibility: Compatible with anthropics/skills and claude-code plugins, allowing immediate use of existing skills and plugins.
- Diverse backends: Supports various LLMs including Anthropic, OpenAI, Gemini, Groq, and Ollama.
- Context management: The Auto-Compaction feature compresses context while preserving task state and logs, enabling long-term session maintenance.
- Safety and collaboration: Provides a multi-level permission system and Swarm Coordination for subagent delegation and management.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
TrueForge Releases Open-Source Harness for Running LLMs as Real Agents
Jevper Released: A Jev Classification Wrapper for OpenAI-Compatible Models with Automatic Logprobs DetectionWhiteboard Launches Open-Source VS Code Fork for Human-AI Collaborative Software DesignOpenComputer Hosts 'Ship Video' Agent Using Claude Opus 5.5 to Generate Explainer Videos from Code