HomeProjectsArticlesPK BoardAbout
Explore Projects

OpenHarness

Normal
GitHub Python MIT

Description

OpenHarness is an open agent harness platform with a built-in personal agent called Ohmo, providing an integrated solution for agent development, testing, and deployment.

Key Features

  • Agent loop with streaming tool-call cycle, API retry with exponential backoff, parallel tool execution, and token counting
  • 43+ built-in tools covering file operations, shell commands, search, web, and MCP server integration
  • On-demand skill loading from markdown files with a plugin ecosystem for skills, hooks, and agents
  • Persistent context and memory via CLAUDE.md discovery, MEMORY.md, context compression, and session resume
  • Multi-level governance with path-level permissions, PreToolUse/PostToolUse hooks, and interactive approval dialogs
  • Swarm coordination for subagent spawning, delegation, team registry, and background task lifecycle

Use Cases

💡 Building production AI agents with a complete infrastructure wrapping an LLM for tool-use, memory, and safety
💡 Rapid prototyping and testing of new agent skills and coordination patterns in a controlled harness
💡 Automating long-running coding tasks like forking branches, writing code, running tests, and opening PRs via ohmo
💡 Researching and extending agent architectures with custom plugins, providers, and domain knowledge

Strengths & Limitations

✅ Strengths

  • • High community interest (15.4k stars)
  • • Permissive open-source license (MIT)
  • • Responsive to issues, low backlog

Tags

framework agent python tools automation

Categories

🤖 Agent Framework

Quick Start

Install OpenHarness via the install script, which links the 'oh' command to ~/.local/bin. Run 'oh' to launch the agent in the terminal with React Ink TUI. Configure authentication with your preferred LLM provider (Claude, OpenAI, etc.). Explore built-in skills with 'oh --dry-run' to preview resolved settings without executing. Start a conversation with ohmo in Feishu, Slack, Telegram, or Discord to begin automating tasks.
Visit GitHub

Related Projects

Atomic Agents

6.2k · Python
Active A+

Atomic Agents is a modular AI agent building framework with an atomic design philosophy, providing composable components including tools, pipelines, and memory management for constructing agent systems.

frameworkpythonagent +2
  • · Atomic design philosophy with single-purpose, reusable, composable components
  • · Built on Instructor and Pydantic for type-safe, predictable agent behavior
  • · Multi-provider support including OpenAI, Anthropic, Groq, Gemini, and more
View full details → View on GitHub

AutoChain

1.9k · Python
Stale C

AutoChain is a lightweight, extensible, and testable LLM Agent framework by Forethought, providing clean abstractions for agent building with automatic tool selection, conversation history management, and automated testing workflows.

frameworkagentpython +3
  • · Lightweight and extensible generative agent pipeline with minimal abstraction layers
  • · Support for custom tools and OpenAI function calling with automatic spec conversion
  • · Simple memory tracking for conversation history and tool outputs with BufferMemory
View on GitHub

Ouroboros

5.5k · Python
Active A

Ouroboros is a spec-driven multi-agent framework that shifts from traditional prompting to specification-driven development, supporting multi-agent collaboration, MCP tool integration, and automated workflow orchestration for building high-quality agent systems.

multi-agentframeworkagent +4
  • · Spec-first workflow: Socratic interview → immutable seed spec → execution → evaluation
  • · Multi-runtime adapter supporting Claude Code, Codex CLI, OpenCode, Hermes, Gemini, Kiro, Copilot
  • · Three-layer architecture: Kernel (contract engine), Plugins (domain workflows), Shell (TUI cockpit)
View full details → View on GitHub

Related Articles

Agent 评估LLM 评测自动化测试

Agent Evaluation and Testing: From Vibe Checks to End-to-End Pipelines

Most teams evaluate agents by checking a few examples. Real evaluation needs layered metrics, non-rotting datasets, and judges that push back. This article provides runnable code patterns and a practical decision framework.

Read more →