HKUDS

HKUDS / OpenHarness

Public

"OpenHarness: Open Agent Harness -- Ultra-Lightweight Claude Code"

47
4
100% credibility
Found Apr 02, 2026 at 106 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

OpenHarness is a lightweight open-source CLI tool that provides an agentic coding assistant with tools, skills, plugins, memory, tasks, and a React terminal UI, compatible with Anthropic models and skills.

How It Works

1
🔍 Discover OpenHarness

You hear about a simple tool that turns AI into a helpful coding buddy, much lighter than big apps.

2
📦 Get it set up

Download and prepare it on your computer with everyday tools like Python.

3
🔗 Link your AI

Tell it which smart service to use by sharing a simple access code.

4
🚀 Start chatting

Type 'oh' to open a friendly chat window where your AI assistant waits to help.

5
💬 Ask for code help

Describe your coding task, and watch it read files, run commands, and suggest fixes.

6
🛠️ Add special skills

Drop in ready-made guides for things like git commits or debugging to make it smarter.

✅ Code like a pro

Your projects get built faster with safe, smart help—everything works smoothly!

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 106 to 47 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is OpenHarness?

OpenHarness is an ultra-lightweight Python harness for turning LLMs into full agent systems, delivering 80% of Claude Code's core functionality with just 3% of the code bloat. Developers get a simple CLI (`oh`) that launches interactive sessions with 43 tools for file ops, bash execution, web search, MCP integration, and multi-agent coordination, plus a React terminal UI for smooth conversations. It solves the problem of heavy, closed agent frameworks by providing open, extensible infrastructure for tools, skills, memory, permissions, and tasks.

Why is it gaining traction?

It stands out as a pure Python drop-in for Claude Code's ecosystem—compatible with official skills and plugins—while slashing complexity from 500k+ lines to 11k, making it dead simple to hack, deploy, or integrate into scripts. The React TUI offers real-time tool feedback, permission prompts, and command pickers (`/` prefix), with non-interactive modes for JSON/stream output in pipelines. Solid test coverage (114 unit + E2E suites) and one-command setup hook devs tired of enterprise overhead.

Who should use this?

AI researchers prototyping agent loops or multi-agent teams without vendor lock-in. CLI power users scripting code reviews, bug fixes, or notebook edits via bash/tools. Python devs building custom agents on top of Anthropic-compatible APIs, especially those extending with skills like commit workflows or PDF processing.

Verdict

Try OpenHarness if you want a lean Claude Code clone—it's battle-tested with passing E2E but early at 47 stars and 1.0% credibility, so expect rough edges in docs and edge cases. Great for experiments, pair with production tools for real workflows.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.