ManzDev

ManzDev / shama

Public

Local Llama.cpp + opencode for development

youtu.beVvb_KYCTD2w
20
0
100% credibility
Found Apr 26, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

SHAMA is a terminal-friendly wrapper that scans local AI model files, lets users select one, and starts a server for easy use with AI development tools.

How It Works

1
🔍 Discover SHAMA

You find a handy tool that makes it simple to run powerful AI brains right on your own computer using your graphics card.

2
🛠️ Prepare your setup

Make sure your computer has the basic helpers like a container runner and graphics drivers ready for AI work.

3
⚙️ Ready the AI core

Build the main AI runner program tailored to your computer's power for the best speed.

4
📥 Gather AI models

Download smart thinking files from a sharing site and place them in your personal models folder.

5
🚀 Add SHAMA helper

Copy the friendly SHAMA script to your tools spot and give it permission to run anytime.

6
⭐ Choose and launch

Open your command window, run SHAMA, pick a model from your list, and start your private AI server.

✅ AI magic unlocked

Your super-smart local AI is now live and ready to team up with your creative projects for instant thinking power.

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 shama?

Shama is a JavaScript-based CLI wrapper that simplifies running llama.cpp servers locally with your GGUF models, scanning your `~/.models/gguf` folder and launching a selected llama.cpp local model on a GPU via Docker or WSL2. It auto-updates your opencode config to point to the local llama.cpp server at `http://127.0.0.1:8999/v1`, letting you use local llama.cpp AI for coding without cloud dependencies. Developers get a clean terminal interface to pick models like Qwen or Gemma and spin up a local RAG-capable llama.cpp setup in seconds.

Why is it gaining traction?

It stands out by bundling llama.cpp local network setup with opencode integration, offering a dead-simple alternative to claude code local llama.cpp or cloud-based GitHub Copilot. The hook is one-command model selection and server launch, perfect for local GitHub actions runner testing or n8n local llama-cpp workflows—no manual config hassles. With NVIDIA GPU passthrough, it delivers responsive local GitHub Copilot alternative performance on decent hardware.

Who should use this?

AI-curious backend devs building local GitHub instances or runners who want offline code completion via opencode local llama cpp. Hardware enthusiasts with RTX cards testing llama.cpp local AI for schamane-style local RAG experiments. Teams ditching SaaS for a local llama cpp server in dev environments.

Verdict

Try shama if you need a quick local llama.cpp + opencode setup, but its 20 stars and 1.0% credibility score signal early-stage maturity with basic docs—expect some setup tweaks for your GPU. Solid for hobbyists, skip for production until it stabilizes.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.