LLM: open-source projects on GitHub
2094 repositories on LLM, picked by the REDDYX AI agent. Each has a summary, key features and a breakdown. Sorted by stars.
// RELATED TOPICS
SELF-HOSTED ORCHESTRATOR FOR AI AGENTS
LoopForge is a self-hosted platform for orchestrating coding agents. Manages CodeBuddy, Codex, and Claude Code via scripts with human approval gates. Supports local workf
★ 850 · GoJADX MCP Server for AI
MCP server enabling the JADX‑AI plugin to integrate large language models for code analysis.
★ 848 · PythonAGENTIC SOC PLATFORM – POWERFUL OPEN
An automated security center with a flexible architecture and agent support, fully open and written in Python.
★ 846 · PythonDEV-PROXY: SIMULATE FAILURES AND LOAD
dev-proxy simulates API failures, rate limiting and chaos from the command line. It lets you configure fault‑injection scenarios and observe client reactions. Ideal for t
★ 846 · C#BAMBOOAI AI DATA ANALYST
BambooAI — AI data analyst that performs transparent reproducible analysis step by step in Python, verifying the evidence behind its conclusions. Supports many models, in
★ 846 · PythonLLM ON PS5: LLAMA.CPP NATIVELY
PS5LM ports llama.cpp to jailbroken PlayStation 5. Runs any GGUF models natively on console, including Qwen 3.8. 668 stars on GitHub, written in C++.
★ 846 · C++MCP Server Octocode
MCP server for semantic code search and real-time context generation using LLM.
★ 845 · TypeScriptLLMCORD: Discord as LLM Interface
Connect any model via OpenAI-compatible API to Discord and chat with AI right in chat.
★ 843 · PythonLLaVA++: NEW LEVEL OF MULTIMODALITY
LLaVA++ adds support for Phi‑3 and LLaMA‑3 models to the base LLaVA, enabling multimodal chat and image generation. Users gain better text‑and‑image understanding thanks
★ 843 · PythonLLM-Engine: Production Scaling
Open-source engine for scaling and deploying large language models with Python support and flexible configuration.
★ 842 · PythonOpen AI Workstation
Open‑WebUI Tools — a set of 15+ tools and filters that turn Open WebUI into a powerful AI workstation.
★ 837 · PythonOMEGA‑AI JAVA FRAMEWORK FOR DEEP LEARNING
Omega-AI - Java framework for deep learning with automatic differentiation, multithreading, and GPU acceleration via CUDA and cuDNN. Enables fast building and training of
★ 837 · JavaTONL - TOKEN OPTIMIZATION IN TYPESCRIPT
TONL is a markup language that optimizes tokens for more efficient LLM operation and reduces query costs.
★ 833 · TypeScriptONE GATEWAY FOR ALL LLMS
respan-gateway — a single endpoint for all LLM providers. Supports any SDK, any model, and tools. One Rust binary replaces dozens of integrations.
★ 833 · RustHELIX: Private AI Agent Fleet
Flagship fleet of AI agents, each gets its own GPU-accelerated desktop. Claude, Codex, Gemini on a private stack.
★ 831 · GoPI-BOLT ACCELERATED AI AGENT
Pi-Bolt is the Pi agent compiled ahead of time to native code. It starts three times faster than Pi on Bun and uses about a third of its CPU.
★ 828 · TypeScriptPULLBOARD AUTONOMOUS DEVELOPMENT
Pullboard — platform for autonomous development of real products. You approve specifications, set rules, and AI‑agents write code and verify each other's work. Nothing re
★ 823 · JavaScriptSCIENCECLAW: AI COLLEAGUE FOR SCIENTISTS
🔬🦞 Self-improving AI assistant for researchers. 285 skills, zero hallucinations, persistent memory.
★ 820 · TypeScriptGeolook: Analysis, Diagnosis, Execution
Open‑source Python geospatial analytics project: from status and diagnostics to creating tasks and their execution.
★ 818 · PythonJAVIS OS: AGENT PLATFORM ON PYTHON
Javis OS — platform for managing AI agents: task orchestration, execution optimization and flexible configuration. Written in Python, supports extensions via API and CLI.
★ 818 · PythonBossConsole - Powerful Console for Agents
BossConsole is an open cross-platform console tool for launching AI agents with a browser, terminal, and 100+ MCP tools.
★ 817 · KotlinSemantic MCP Server
The MCP server provides real-time semantic code search and context generation with LLM, working with public and private repositories.
★ 816 · TypeScriptAI CHECKS CODE BY BOOKS
AI code review based on 12 classic engineering books with risk diagnosis, quotes, and auto‑fix. 815 stars!
★ 815 · JavaScriptRISC-V CPU COMPILED INTO QWEN3.5 WEIGHTS
Qwen3.5-4B with a 32-bit RISC-V CPU compiled directly into model weights. CPU uses Gated DeltaNet state as RAM and executes code inside model forward pass.
★ 814 · PythonTOKENTAP — CATCHES ALL LLM REQUESTS
Intercept LLM API traffic and visualize token usage in real time. Track costs, debug prompts, and monitor the context window.
★ 812 · PythonGHIDRA + MCP New Reverse
Native MCP server for Ghidra, enabling AI assistants to analyze binaries and automate reverse engineering.
★ 811 · JavaACADEMY AGENT: COMMERCIALIZATION
Turn any scientific article into a commercialization report: 6 AI agents, TRL/MRL, patent landscape.
★ 811 · PythonUI COMPONENTS FOR AI AGENTS
affectui — React components for AI agent UI: trees, tool calls, traces, handoffs and approvals. Install via shadcn or MCP — one file per component. Open source, 376 stars
★ 809 · TypeScriptRust + LLM: Indexing Speed
Swiftide — fast streaming indexing and queries for AI agents on Rust. 710⭐ and growing.
★ 806 · RustCoreCoder: Minimal AI Agent
A compact AI agent for programming (~1400 lines of Python), inspired by Claude Code. Works with any LLM.
★ 804 · PythonThree Agents — Architect, Builder, Reviewer
Three AI agents: architect designs, builder writes code, reviewer checks. Token‑optimized, works in Claude Code, VS Code, Cursor and any IDE with context files.
★ 804 · ShellMASCOT: Multi-Agent Assistant
MASCOT is a multi-agent socially collaborative assistant presented at EMNLP 2026. The project is implemented in Python and has already garnered 518★.
★ 804 · PythonGPT-VIS: AI DATA VISUALIZATION
🤖 GPT-Vis — data visualization for AI Friendly! AntV's entry for AI, 100% generated by AI. Explore data in a convenient visual format.
★ 802 · TypeScriptDIGITAL AGENT SYRENE
Cyrene-Agent — an open‑source desktop AI agent with Cyrene’s persona, built on the Cyrene_Harness framework. It blends lifelike companion chat with practical features: co
★ 801 · TypeScriptSWIFTLM: SWIFT LLM ON APPLE
Native LLM inference server in Swift with MLX, OpenAI‑compatible API, SSD streaming for MoE >100B and TurboQuant KV‑cache compression.
★ 799 · SwiftFABLECUT VIDEO EDITOR FOR AI
Zero‑dependency browser video editor controlled by AI agents via JSON timeline and MCP/REST. Live UI reload makes editing instantaneous.
★ 797 · JavaScriptSURROGATE — POWERFUL FINE-TUNING IN C++
Training and fine-tuning models at the speed of light. 796 stars on GitHub.
★ 796 · C++PDF_OXIDE — FASTEST PDF
The fastest PDF library for Python and Rust. Extract text/images, convert to markdown, create and edit.
★ 793 · RustAEON: Autonomous Agent Framework
AEON is a framework for fully autonomous AI agents without approval loops or constant supervision. Set up once and forget.
★ 793 · TypeScriptPARROT.NVIM: STOCHASTIC PARROT
The plugin adds an AI parrot to Neovim, generating code and comments on the fly.
★ 792 · LuaAGENTSIAGHT Kernel Tracing
agentsight – zero‑instrument system‑level AI agent tracing via eBPF. Allows tracking agent behavior without code changes.
★ 791 · CALPHA AVATAR: LIVING OMNI-AVATAR
Real-time interactive avatar on LiveKit, compatible with any open-source modules: model, voice, vision, memory, and search.
★ 790 · PythonLLM-APPS-JAVA-SPRING-AI — QUICK AI START
Examples of Java applications on Spring Boot using Spring AI and LLM for generative AI.
★ 789 · JavaClaude-TAP: Agent Traffic
Intercept and analyze API traffic from Claude Code, Codex CLI, Gemini CLI, Cursor CLI, OpenCode, Kimi, Pi, and Hermes in a local trace viewer.
★ 788 · PythonTRACK OPENSWARM PROJECT
Open control libraries; AI development playlist
★ 788 · TypeScriptTERMINAL AGENT FOR DEEPSEEK V4
An economical terminal agent with self-verifying memory and support for complex long tasks.
★ 787 · TypeScriptTURN ANY REPO INTO RL ENV
Repo2RLEnv turns any GitHub repository into a verifiable RL environment for coding agents. Harbour tasks prepare, evaluate, and publish to Hugging Face Hub. 739 stars, Py
★ 787 · PythonAFFILIATE-SKILLS 50 AI SKILLS
A set of 50 AI agent skills for affiliate marketing: trend research, post writing, infographics, landing pages, and deployment.
★ 786 · HTMLMULTI-AGENT FRAMEWORK ON RUST
AutoAgents — a Rust framework for creating, deploying, and coordinating multiple intelligent agents in a single system.
★ 784 · RustTHOUGHTDAG Mind Map Online
Infinite canvas for LLM dialogues: turn conversations into an editable thought graph, where connections are context.
★ 783 · TypeScriptDISTILL: LIGHTWEIGHT CODING AGENT WITH TUI
Distill is a lightweight coding agent with a terminal interface, written in Rust. It enables achieving greater results by using significantly fewer tokens when working wi
★ 781 · RustFree LLM API Keys
Free API keys for GPT-5.4, Claude, DeepSeek, Gemini, Grok — copy, paste, use. Updated 3-5 times a day; no credit card required.
★ 780 · PythonCACHE-TO-CACHE: DIRECT LLM EXCHANGE
Implementation of the Cache-to-Cache algorithm for direct semantic exchange between large language models. Enables transferring KV‑cache without deserialization speeding
★ 778 · PythonDEEPZERO: FIND VULNERABILITIES WHILE YOU SLEEP
An automated framework finds vulnerabilities in Windows drivers by decompiling and analyzing IOCTL with AI agents.
★ 777 · PythonTRINITY: PLATFORM FOR AI AGENTS
Trinity is a self‑hosted platform for running AI agents Claude Code, Codex and Gemini. It provides orchestration, versioning and access‑policy management. Licensed under
★ 775 · PythonDAPR-AGENTS: DURABLE AI AGENTS
Framework for building autonomous agents with built-in workflow orchestration, state, and telemetry. Ensures resilience and observability in production.
★ 774 · PythonSoduWork: ENGINE FOR AI AGENTS
AI-native platform for managing agent teams. Supports multi-harness approach, workflow orchestration, and self-hosted deployment.
★ 774 · JavaScriptRECALL — Memory for Claude Code
Recall gives Claude Code long-term memory — fully offline. No need to explain the project again in every session.
★ 769 · PythonDINGO: AI MODEL EVALUATION
Dingo is a tool for comprehensive evaluation of data, model, and AI application quality in Python. It enables measuring reliability and performance of systems.
★ 768 · PythonLOCAL-RAG: RAG WITHOUT DATA LEAK
Ingest files for RAG with an open-source LLM — everything runs locally, with no third‑party services and no data leaving your network.
★ 768 · PythonNew projects on this topic in Telegram
Telegram →