Explore the directory

A modular framework for post-training large language models, providing unified GPU backends and system optimizations for reinforcement learning that plug into existing training frameworks.

• Models

A structured, searchable learning path covering retrieval-augmented generation, embeddings, vector search, agents and Model Context Protocol, aimed at engineers who build with LLM APIs rather than train models.

• Tutorials
• Books

A federated search index over Agentic Resource Discovery registries, letting an AI agent find MCP servers, APIs, skills and other tools by query instead of being wired to each one in advance.

• Agent frameworks
• APIs

A Claude Code skill that reads a GitHub repository and produces a live explainer web page with architecture and data-flow diagrams, along with a downloadable vector knowledge base, a search command-line tool and an MCP server built from the repository's own source.

• Agent skills
• Documentation tools

A native C and Metal inference engine that runs large mixture-of-experts language models, such as a 284-billion-parameter DeepSeek model, on a single Apple Silicon Mac by streaming quantized experts from disk instead of holding everything in memory.

• Inference

A step-by-step, freely available class that teaches data analysis with Python and Jupyter notebooks, aimed at journalists and other newcomers to programming.

• Tutorials

A PyTorch runtime for full-parameter fine-tuning of large language models on a single GPU, which streams inactive layers between GPU memory, CPU RAM and optional NVMe storage.

• Models

Fast-booting MicroVM sandboxes for AI agents, with warm pools for near-instant claims, a per-node control plane, and Go and Python SDKs for executing commands and files inside the guest.

• Containers

A command-line tool that compiles one shared AI assistant configuration into the separate formats used by Claude Code, Cursor, GitHub Copilot and other coding assistants, flagging drift.

• Agent skills
• CLI tools

A Python library that integrates embedding and reranker models from providers such as OpenAI, Voyage AI, Cohere and Hugging Face for use with the Milvus vector database.

• RAG
• Databases

A curated list of Model Context Protocol clients, server implementations and frameworks, organized by category such as file systems, databases, communication and developer tools.

• MCP servers
• Curated lists

A containerized AI coding workstation that bundles OpenCode with more than fifty developer tools, a sandboxed headless browser and support for over ten AI providers, including a user's own Claude subscription.

• Coding agents
• Containers

A fail-closed execution firewall that sits between AI coding agents and the system, allowing, denying or escalating each action and writing a signed, offline-verifiable receipt of the decision.

• AI evals
• Security

An open-source deep-research MCP server that fuses results from several search sources into a cited synthesis, for Claude Code, Codex, Cursor and other Model Context Protocol agents.

• MCP servers
• RAG

Rust bindings for Meta's ExecuTorch library, letting PyTorch models exported in Python be loaded and run on mobile, embedded and edge devices from Rust code.

• Inference
• Mobile

A tool that installs an offline AI coding agent onto a USB drive, so plugging it into a Windows, macOS or Linux laptop gives a terminal coding agent backed by a local model with no install on the host.

• Coding agents

A lightweight persistent-memory system for Claude Code that captures coding decisions and bug fixes across sessions through an MCP server and hooks, using full-text search instead of a vector database.

• MCP servers
• Agent memory

A curated collection of over two hundred deep-dive tutorials that explain the internal architecture of AI agents, LLM frameworks, coding tools and other open-source developer platforms.

• Tutorials
• Curated lists

A statically typed programming language designed for AI models to write and humans to review, with a Lean-checked certificate of proof obligations attached to compiled code.

• Languages

An open-source, self-hosted platform for orchestrating autonomous multi-agent AI systems, with a visual workflow builder, human-in-the-loop approvals and many Model Context Protocol tools.

• Agent frameworks
• Workflow automation

A Codex skill and bundled command-line client that edits and manages Overleaf LaTeX projects remotely, sending small verified changes to specific documents instead of downloading and re-uploading the whole project.

• Agent skills
• CLI tools

A native macOS and iOS app that runs open-weight language models far larger than the device's memory by streaming each layer's weights from an external SSD within a memory budget the user sets, entirely on-device.

• Inference

A community-maintained knowledge base of ClickHouse solutions and troubleshooting notes, written by Altinity engineers and the ClickHouse community and published as a searchable website.

• Databases
• Documentation tools

A Claude Code hook that intercepts build commands such as cargo or gcc and transparently runs them on remote worker machines, so concurrent AI coding agents do not saturate one workstation's CPU.

• Agent skills
• Hosting