LiteLLM-Labs

LiteLLM-Labs / lite-harness

Public

Unified Server for running OpenCode, Claude Code, Codex agents

16
3
85% credibility
Found May 31, 2026 at 16 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

LiteLLM Harness is a unified server that lets teams deploy and manage AI coding agents (Claude Code, GitHub Copilot, OpenCode, Codex) through a web interface, command line, or Slack, with secure secret storage, reusable skills, scheduling, and human approval workflows.

How It Works

1
🚀 Install the platform

You add the platform to your AI coding assistant with one simple command, and everything connects automatically.

2
🔑 Connect your AI service

You enter your AI gateway address and key so your agents can think and reason using your preferred AI models.

3
🗄️ Store your secrets safely

You save API keys and credentials in the encrypted vault — your agents can use them without ever exposing them publicly.

4
🤖 Create your first agent

You define what your agent should do: write a description, set instructions, attach useful skills, and choose when it should run.

5
Choose how to chat
🌐
Web browser

Use the polished web interface to chat, review code, and manage everything visually

⌨️
Terminal

Chat directly from your command line for a fast, keyboard-driven experience

💬
Slack

Message your agent right from Slack and it responds as a bot in your workspace

6
✅ Agent works autonomously

Your agent runs tasks in a secure sandbox, asks for your approval when needed, and reports back when it's done.

🎉 Your team has AI agents

Everyone on your team can now use the same powerful AI coding agents through a single, unified server — scheduled or on-demand.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 16 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 lite-harness?

Lite-harness is a unified server that lets you run any AI coding agent -- Claude Code, Codex, OpenCode, or GitHub Copilot -- behind a single API. Instead of managing separate services for each agent with different configs, you get one Docker container that fronts them all. Agents talk to your LiteLLM gateway, run in isolated sandboxes, can be scheduled via cron, and surface approval requests to humans through a web inbox or Slack. The frontend is a Next.js app; the CLI gives you interactive chat sessions; the API speaks OpenCode's spec.

Why is it gaining traction?

The hook is consolidation. If you're running opencode in one place, claude-code in another, and stitching them together with custom tooling, this eliminates that mess. One service, shared MCP tools, shared vault secrets, shared session management. The Slack integration built into the agent workflow is also a real differentiator -- agents can post to channels and respond to mentions without extra infrastructure. The vault system for injecting credentials into sandboxes hit the right abstraction level for teams. And the human-in-the-loop approval flow through a dedicated inbox treats agents as actual team members rather than black boxes.

Who should use this?

Platform and DevOps teams tired of juggling multiple agent runtimes. Organizations running scheduled agent workflows that need centralized credential storage and audit trails. Any team that wants AI agents accessible through Slack without building custom integrations. If you're a solo dev running one agent locally, this is overkill -- but if you're an eng team trying to operationalize agent work across the organization, the unified surface is the draw.

Verdict

The credibility score (0.85 per Weave) reflects a well-structured, thoughtful codebase, but 16 stars means the project is early and community proof is thin. Docs are solid, the CLI works, and the Docker setup is production-ready. The LiteLLM dependency is a requirement not a nice-to-have, so factor that in. Worth piloting if your team juggles multiple agent runtimes, but treat it as cutting-edge tooling rather than a stable enterprise stack.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.