Directory
Explore the directory
Tools, libraries, services and references, each reviewed before it goes public. 6,821 resources listed.
MindMemOS
github.comA portable, self-evolving memory layer for AI agents that stores user and task context and reuses it across agents, automatically distilling recurring experience into reusable skills.
RhinoMCP
github.comRhinoMCP connects Rhino 3D and Grasshopper to AI agents through MCP, so an assistant such as Claude or Cursor can model geometry, read the current document, capture the viewport for visual feedback, and build or edit Grasshopper definitions from plain language.
Awesome OpenClaw
github.comA curated list of resources, tools, skills, tutorials and community projects for OpenClaw, the open-source self-hosted AI agent for WhatsApp, Telegram, Discord and other channels.
Microsoft 365 MCP Server
github.comThe Microsoft 365 MCP Server is a Model Context Protocol server that exposes more than 300 tools over the Microsoft Graph API, covering Outlook mail and calendar, OneDrive, Excel, OneNote, To Do, Planner and Contacts, authenticated through the Microsoft Authentication Library.
onnx2tf
github.comonnx2tf converts ONNX model files to LiteRT, TensorFlow Lite, TensorFlow, native PyTorch modules, TorchScript, PyTorch state dictionaries and exported PyTorch programs, and can also convert directly from LiteRT back to PyTorch.
Learn Agent Architecture
github.comLearn Agent Architecture is a section-by-section guide to how the harness around a language model turns reasoning into controlled action: the agent loop, tools, memory, permissions and context, illustrated with real systems such as Claude Code and Hermes Agent.
A curated collection of papers, tutorials, books, benchmarks and open-source libraries on out-of-distribution detection, robustness and generalization in machine learning.
GraphRAG-SDK
falkordb.comGraphRAG-SDK is a GraphRAG framework built on the FalkorDB graph database that grounds LLM answers in a knowledge graph rather than vector similarity alone, tracing each answer to its source chunks so an app can inspect the evidence or decline to answer.
ByteChef
bytechef.ioByteChef is an open-source, Apache-2.0 platform for AI agents and workflow automation, positioned as an alternative to n8n and Zapier that can be self-hosted or embedded, with a drag-and-drop AI Agent step that runs the full model-and-tool loop in a durable workflow.
FlashRank
github.comFlashRank is an ultra-lightweight Python library that reranks search results before they are fed to an LLM, using cross-encoder and listwise LLM-based rerankers small enough to run on a CPU with no PyTorch or Transformers dependency.
Parallel Code
parallelcode.appParallel Code is a free, open-source desktop app that runs Claude Code, Codex and Gemini side by side, each isolated in its own git worktree and branch, so several coding agents can work on different tasks at once without touching each other's changes.
Multi-Agent CAD (MAC)
github.comMAC is a multi-agent framework turning a natural-language design brief into editable CAD geometry, generating a single printable part or decomposing a mechanism into separately generated components that it assembles, validates and exports for inspection or simulation.
Colab-mcp
github.comColab-mcp is a Model Context Protocol server that bridges a local coding agent to a Google Colab notebook session running in the browser, so a client such as Gemini CLI, Claude Code or Windsurf can work with a live Colab session.
Atlassian MCP Server
atlassian.comThe Atlassian MCP Server is Atlassian's own cloud-hosted MCP server giving AI tools such as Claude, ChatGPT, Cursor and VS Code secure, real-time access to a user's Jira, Confluence, Service Management, Bitbucket and Loom data through OAuth 2.1 or API tokens.
Naersk
github.comA Nix library that builds Rust projects into Nix derivations directly from Cargo.lock, without extra configuration, code generation or import-from-derivation.
Awesome DSH Plugin
github.comA curated guide to DeepSeek Harness, DeepSeek's open-source everything-is-a-plugin coding agent, and its community plugin ecosystem.
ECA (Editor Code Assistant)
eca.devECA is an editor-agnostic, LSP-inspired protocol and server, written in Clojure, for AI pair programming: it gives chat, rewrite and completion the same behavior in Emacs, VS Code, IntelliJ, Neovim or a desktop client, all configured once regardless of which editor is running it.
A bilingual web app that lets a user drag a camera in 3D space or use sliders to set azimuth, elevation and distance, then generates a new image of an uploaded photo from that viewpoint.
workflow_templates
pypi.orgThe official collection of ComfyUI workflow templates and reusable subgraph node blueprints, distributed as Python packages and shown on a searchable template site.
FastEmbed-rs
docs.rsFastEmbed-rs is a Rust library for generating text and image vector embeddings and for reranking search results locally, using ONNX Runtime for inference and Hugging Face tokenizers for encoding, with synchronous use and no dependency on Tokio.
ClawKeeper
github.comClawKeeper is host-agnostic safety middleware between an AI agent and its tools, blocking risky shell, Python, browser, filesystem and network calls, redacting sensitive results, and escalating harder decisions to a Watcher that learns new guard patterns.
Awesome DevOps MCP Servers
github.comA curated list of Model Context Protocol servers for DevOps work, covering infrastructure-as-code, cloud platforms, CI/CD and other operational tooling, so an AI assistant can act on real infrastructure through a standardized server interface instead of a one-off integration.
This project studies why on-policy distillation of large language models succeeds or fails: it finds the technique needs a student and teacher with compatible thinking patterns and genuinely new capability, then proposes two ways to recover a failing run.
AI Agents from Scratch
github.comAI Agents from Scratch is a 12-lesson, local-first course that builds one AI agent step by step from a single local LLM call, with no frameworks and no cloud APIs, so a developer can see exactly how the loop, tools, memory and planning underneath an agent actually work.