#gguf

Blink - Local AI Code Completions

Lojell

Lightning-fast inline AI completions (ghost text), local-first and BYOK: GGUF models in-process via llama.cpp, or any OpenAI-compatible endpoint.

DeskAI — Local AI Coding Assistant

AbhishekMishra9454

100% offline AI coding assistant. Full project generation, agentic file editing & terminal control powered by local LLMs (Qwen3, DeepSeek-Coder) running entirely on your machine. Works on Windows, macOS and Linux. No API keys.

Forge LLM

Efsoo

Local coding agent with first-class llama.cpp/GGUF control, reliable tools, and Keep/Undo checkpoints

Weightless

kipngenokoech

Inspect models without loading the weights. Architecture flowcharts, tensor tables, backbone detection, reconstructed PyTorch definitions, memory & finetune estimates — header-only reads open multi-GB .safetensors and .gguf files instantly, even straight from the Hugging Face Hub.

siGit Code — On-device AI Coding Agent

getsigit

Local-first AI coding agent for VS Code. Runs the on-device sigit agent, and other ACP-compatible agents, over stdio.

Trie Coding Agent

Trie

Coding agent: local GGUF via embedded trie-daemon, or any LLM API (Ollama, LM Studio, OpenAI, Kimi, and other cloud endpoints). Hybrid mode adds sparing frontier-model guidance.

Vectra

laudarisd

A local-first AI coding agent for VS Code. Run GGUF models with llama.cpp, understand repositories, review edits, and execute approved tools.