zhijiewong

zhijiewong / openharness

Public

Open-source agent harness framework — build your own terminal cli with any LLM

20
3
100% credibility
Found Apr 01, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

OpenHarness is an open-source terminal-based AI coding agent that works with local or cloud language models to read, edit files, run shell commands, and assist with development tasks through conversational interaction.

How It Works

1
📰 Discover OpenHarness

You hear about this friendly coding helper that chats with you right in your terminal to build and fix code.

2
📥 Get it running

You install it quickly and launch with a simple command, seeing a fun banner appear.

3
Pick your thinking partner
🆓
Local free

Use what's already on your machine – no extra setup, fully private.

☁️
Cloud smart

Link a service so it taps into bigger brains when needed.

4
💬 Start chatting

Type what you need like 'fix this bug' or 'add login page' and watch it respond step by step.

5
✅ Approve actions

It asks permission before reading files, making changes, or running tests – you stay in control.

6
✨ See the magic

It reads your code, suggests edits, runs checks, and updates everything perfectly.

🎉 Code complete

Your project is improved, bugs gone, new features working – coding feels easy and fun.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 20 to 20 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 a TypeScript CLI tool that transforms any LLM into a terminal-based coding agent, mimicking Claude Code but without vendor lock-in. Run `oh` in any project to chat, letting the AI read/edit files, run bash commands, grep code, glob files, or fetch web content via built-in tools. It auto-detects your project (TypeScript, Python, Rust), loads `.oh/RULES.md` for guidelines, and supports local Ollama models for free, offline use or cloud APIs like OpenAI and Anthropic.

Why is it gaining traction?

As an open source agent CLI and GitHub Copilot alternative, it stands out by working with any LLM—local or cloud—via simple flags like `oh --model ollama/qwen2.5:7b`. Permission modes (ask/trust/deny) prevent rogue edits, tools run safely with concurrency for reads, and commands like `oh init`, `oh sessions`, or `oh models` make workflows snappy. Developers dig the React-powered Ink UI for real-time streaming in the terminal, plus session resuming and cost tracking.

Who should use this?

Fullstack devs debugging Node.js or Python repos locally without API costs. Indie hackers building agentic coding prototypes or self-hosted GitHub alternatives. Teams enforcing rules in open source agent workflows, like always running tests post-edits.

Verdict

Promising alpha open-source agent framework at 18 stars and 1.0% credibility—docs are crisp, install is one-liner, but low adoption means test it for edge cases. Grab it if you want a free, local open source GitHub Copilot alternative; otherwise, wait for more polish.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.