AI tools, MCP servers, LLM frameworks, and agent platforms for developers.
26 posts
Prime Agent is a free, open source self-improving RLM agent for coding and long-running autonomous tasks. Persistent IPython, built-in subagents, continual harness refinement, daemon-backed sessions. 15.8K stars.
NVIDIA Switchyard is a free, open source Rust proxy that routes LLM requests across providers. Translates between OpenAI and Anthropic APIs, supports Claude Code and Codex, A/B benchmarking, cost optimization. Apache 2.0.
Needle 2 is a free, open source 45M-parameter foundation model for tool calling on tiny devices. 14MB binary, 28MB RAM, runs fully offline. LoRA fine-tuning, confidence gating, structured extraction. 5.4K stars.
LoopX is a free, open source state kernel for long-running AI agent teams. Durable goals, quota-aware scheduling, evidence logs, verifiable handoffs. Works with Codex, Claude Code, Cursor. 4.7K stars.
Cloudflare Computer is a free, open source virtual filesystem that lives inside Durable Objects. Give AI agents a real computer with container, shell, and JS execution backends. MCP server included. 8K+ stars.
TencentDB Agent Memory is a free, open source team-level memory hub for AI agents. Turns conversations, docs, and code into 4 reusable memory assets. Chat Memory, Skills, Wiki, CodeGraph. 18K+ stars.
Soup CLI is a free, open source LoRA fine-tuning tool that streams frozen base model layers from RAM to GPU one at a time. Train Llama-3.1-8B at 119.6 tok/s in 3.32GB peak VRAM. SFT, DPO, GRPO, KTO.
reverse-skill is a free, open source cybersecurity skill router for AI agents. 41 routing rules, 163 benchmark cases, supports APK, binary, JS, CTF, pentesting. Works with Claude Code, Cursor, Cline. 22K+ stars.
book-to-skill is a free, open source tool that converts any PDF book into a structured agent skill. Works with Claude Code, GitHub Copilot CLI, and Amp. On-demand chapter loading, 24x-51x fewer tokens than dumping the PDF.
AirLLM lets you run 70B parameter LLMs on a single 4GB GPU without quantization or pruning. Stream layers from disk, one at a time. Free, open source, supports Llama, Qwen, DeepSeek, and Kimi K3.
VoxCPM2 is a free, open source, tokenizer-free text-to-speech and voice cloning engine from OpenBMB that invents new voices from a text description, clones a voice from a few seconds of audio, and speaks 30 languages locally on your own GPU.
MiroFish is a free, open source multi-agent AI prediction engine that spawns hundreds of autonomous LLM agents on a simulated social network to forecast public reaction to news, product launches, and policy drafts before you publish.
AionUi is a free, open source desktop app that wraps Claude Code, Gemini CLI, OpenCode, and 20+ other AI agents in a single graphical workspace, with local SQLite storage, MCP sharing, offline Ollama support, and scheduled automation.
STORM is a free, open source autonomous research system from Stanford that generates full-length, citation-grounded Wikipedia-style articles by simulating multi-perspective expert interviews.
VideoAgent is a free, open source multi-agent AI framework from HKUDS that orchestrates over 30 specialized agents to edit, analyze, and assemble long-form video from natural language prompts.
Microsoft BitNet is a free, open source 1.58-bit ternary weight inference framework that runs massive LLMs on standard consumer CPUs with no GPU required. Complete setup guide for bitnet.cpp.
JARVIS (HuggingGPT) is a free, open source AI system from Microsoft Research that uses an LLM as a brain to automatically select and orchestrate thousands of Hugging Face models for multi-modal tasks.
Heym is a free, self-hosted AI-native workflow automation platform with a visual canvas for multi-agent orchestration, RAG pipelines, LLM tracing, and built-in Model Context Protocol support.
nanoGPT is the simplest, fastest open source framework for training and fine-tuning GPT models. Two files, 300 lines each, pure PyTorch. Free, hardware-agnostic, Flash Attention built in.
MiniMind is a free, open source educational framework that lets you train a GPT-style LLM from scratch in 2 hours on one GPU for $0.50. Pure PyTorch, no black boxes, full pipeline from tokenizer to RLHF.
Osaurus is a free, open source macOS AI harness built in Swift for Apple Silicon. Run local models, route to cloud APIs, and build autonomous agents with persistent memory. No subscription.
OpenDataLoader PDF is a free, open source document parser that preserves reading order, tables, and layouts for AI RAG pipelines. Local AI vision fallback, 93% table accuracy, no cloud required.
Meetily is a free, open source, privacy-first meeting assistant that transcribes and summarizes calls locally on your machine. No cloud, no bots joining your calls, no monthly fees.
Llamafile bundles the inference engine, web UI, API server, and model weights into one portable executable that runs on Windows, Mac, and Linux. Free, open source, zero install.
Hermes Browser Extension is a free, open source browser side panel that pipes live web context into your local Hermes Agent runtime. No cloud, no third-party telemetry, full MCP tool support.
Alook is a free, open source multi-agent orchestration platform that gives AI agents their own emails, kanban boards, and calendars. Self-hosted, private, and 100% free.