ChatGH

tirth8205 / code-review-graph

A local-first code intelligence graph for AI coding tools. Builds a persistent structural map of your codebase via Tree-sitter AST parsing, tracks changes incrementally, and exposes 30 MCP tools for blast-radius analysis, risk-scored PR review, and token-efficient context retrieval. Benchmarked at ~82x median token reduction across 6 real repositories. Supports 35+ languages and integrates with Claude Code, Cursor, Codex, Copilot, and 10+ other platforms.

1jehuang / jcode

A next-generation, Rust-powered coding agent harness built for extreme performance and multi-session workflows. Features swarm coordination for parallel agent collaboration, semantic memory with graph-based recall, 30+ provider integrations, browser automation, and self-modifying capabilities — all in a sub-30MB, sub-15ms TUI.

diegosouzapw / OmniRoute

Open-source AI gateway that unifies 236+ LLM providers into one endpoint. Features 17 routing strategies with 4-tier auto-fallback, RTK+Caveman stacked compression saving 15-95% tokens, MCP/A2A agent protocols, and works with Claude Code, Cursor, Copilot and 16+ coding agents. 50+ free tiers included, local-first and private.

rohitg00 / ai-engineering-from-scratch

A free, MIT-licensed AI engineering curriculum with 435 lessons across 20 phases (~320 hours). Covers math foundations through autonomous multi-agent systems in Python, TypeScript, Rust, and Julia. Every algorithm is built from raw math first, then reinforced with production frameworks. Each lesson ships a reusable artifact — prompt, skill, agent, or MCP server — that plugs directly into your daily workflow.

msitarzewski / agency-agents

A comprehensive collection of ready-to-use AI agent templates organized by professional domains like engineering, marketing, design, and finance. Each agent includes structured prompts, defined workflows, and clear deliverables, enabling developers and teams to quickly assemble specialized AI workforces or multi-agent systems. Includes integration guides for popular AI coding assistants.

kvcache-ai / ktransformers

A CPU-GPU heterogeneous computing framework for efficient LLM inference and fine-tuning. Optimized for Mixture-of-Experts (MoE) models, it enables running ultra-large models like DeepSeek-R1 on consumer hardware with 24GB VRAM. Features Intel AMX/AVX acceleration, INT4/INT8/FP8 quantization, and seamless integration with SGLang and LLaMA-Factory for production serving and fine-tuning.

jamiepine / voicebox

A local-first open-source AI voice studio covering the full voice I/O loop — clone voices, generate speech in 23 languages across 7 TTS engines, dictate into any app via global hotkey, and give MCP-aware AI agents a voice of your choosing. All data stays on your machine with native Tauri performance.

topoteretes / cognee

Cognee is an open-source knowledge engine that transforms documents into AI memory infrastructure. It combines vector search, graph databases, and cognitive science to provide contextual understanding for AI agents. Features include unified ingestion, persistent learning agents, and reliable knowledge management.

Robbyant / lingbot-map

A feed-forward 3D foundation model for streaming scene reconstruction from video sequences. Based on the Geometric Context Transformer architecture, it unifies coordinate grounding, dense geometry, and long-range drift correction with paged KV-cache attention, achieving ~20 FPS inference on ultra-long sequences (10K+ frames). Delivers state-of-the-art results across 10+ benchmarks including KITTI, ETH3D, Oxford Spires, and Tanks & Temples.