CubeCL is a Rust procedural‑macro‑based language that lets you write a single compute kernel in Rust and JIT‑compile it to CUDA, HIP, Metal, WebGPU (WGSL/SPIR‑V) or CPU SIMD. It models parallelism with four axes (vector, plane, cube dimension, cube count), supports compile‑time specialization, automatic vectorization, and runtime autotuning, and forms the low‑level foundation for the Rust deep‑learning framework Burn and the kernel library cubek.

A mobile control interface for the Pi coding agent that lets users chat with and manage their local AI agent remotely via a secure relay system.

A gamified browser extension that unifies user script management and web capabilities like AI script generation, ad-blocking, and video enhancement through a card-based interface.

An agentic AI development platform that helps developers plan, build, and review software by turning prompts into executable specs and using parallel agents across multiple interfaces.

A free and open-source motion capture platform that provides research-grade tracking without requiring expensive, proprietary hardware.

GitHub Agentic Workflows (gh-aw) is an official GitHub CLI extension that lets developers define AI-powered repository automation in Markdown with YAML frontmatter, then compile and run those agentic workflows securely through GitHub Actions.

A provider-agnostic terminal AI coding agent that uses a Think-Act-Verify loop to read, edit, and execute code using any OpenAI-compatible LLM.

An AI director system that automates the entire creative video production pipeline from scriptwriting and character design to storyboarding and final editing.

A self-hosted, no-code canvas for building AI agent workflows that integrates with local or cloud LLMs and over 30 external services.

A plugin for DSH that converts AI model replies into interactive, safe, and visually rich UI components like dashboards and charts.

A security-first MCP server that provides LLM agents with controlled SSH access to remote hosts through command classification, policy-based authorization, and human-in-the-loop approval.

A multi-agent collaboration architecture for intelligent penetration testing and CTFs that uses a Manager, Solver, and Observer system to automate complex attack chains.

WhisperJAV is a locally‑run subtitle generator for Japanese adult videos. It mitigates Whisper’s known failures on noisy, long‑form audio by using scene‑based segmentation, voice‑activity detection, defensive decoding, and Japanese‑specific post‑processing. The tool offers a GUI and a CLI, multiple processing modes (balanced, fidelity, fast, qwen, anime‑whisper, etc.), and an optional two‑pass ensemble that merges results from different pipelines. New in v1.9.2 are run‑summary tables, offline‑only mode, VAD version selection, and better failure handling.

TheRock is the open-source build and release system for HIP and ROCm, providing a modular workflow to integrate and distribute ROCm components.

Vexa is an open‑source platform that lets a bot join Google Meet, Teams, Zoom (and Jitsi) calls, stream speaker‑attributed transcripts in real time, and store them as Markdown in a Git repo you control. Sandbox‑run LLM agents can then read/write that repo to generate summaries, structured wiki pages, and other knowledge artifacts. You can use the hosted service or self‑host the full stack via Docker or Kubernetes, keeping all data on‑premise and bringing your own speech‑to‑text and LLM models.

An Obsidian plugin that runs a local HTTPS REST API and a Model‑Context‑Protocol server, letting scripts, browser extensions, and AI agents perform full CRUD, targeted markdown patches, searches, command execution, and more on your vault. Authentication is via a bearer token and a self‑signed certificate; the plugin also provides ready‑made MCP configs for Claude and Cursor.

Perplexity‑AI is an unofficial Python wrapper that scrapes the public Perplexity.ai web UI to run search, reasoning, and research queries without an API key. It offers synchronous and asynchronous clients, streaming answers, file uploads, and a small MCP server for integration with Claude Code. An optional driver can automate Chrome to create temporary Gmail accounts via Emailnator, letting users refresh the free‑tier “5 pro queries” limit. The library includes typed interfaces, retry/back‑off, rate‑limit handling, and custom exceptions.

Box is an Android app that forks Google’s AI Edge Gallery to provide a comprehensive, offline‑first AI suite. It runs large language models, vision models, diffusion image generators, speech‑to‑text, text‑to‑speech, music generation, and a spoken‑camera assistant (Box Assist) directly on the phone, leveraging CPU, GPU, or NPU acceleration. The app removes Google telemetry, adds biometric lock and encrypted storage, and offers multilingual UI and accessibility features. Installable via APK (standard or custom‑ROM build) and licensed under Apache 2.0.

A collection of samples and tutorials for Amazon Bedrock AgentCore, providing the infrastructure to deploy and operate framework- and model-agnostic AI agents securely at scale on AWS.

Frame is an Electron‑based IDE that gives AI coding agents a persistent, git‑anchored context. It creates a standard `.frame/` folder with rules, a module map, notes, tasks and markdown‑based specs (spec/plan/tasks/outcome). Agents start each session with this context, and a built‑in conductor can run multiple specs in parallel isolated git worktrees, enforcing non‑overlapping footprints and requiring human approval before merging. Features include a multi‑terminal grid, task/spec panels, GitHub integration, fast intent‑index lookup, and support for Claude Code, Codex CLI, and Gemini CLI. Installation is via `npm install` or pre‑built binaries; the project is Apache‑2.0 licensed.

A governance standard and toolset for knowledge repositories maintained by LLM agents, ensuring deterministic work, resumability, and clear operator-agent boundaries.

A TypeScript-based coding agent implementation and interactive tutorial that teaches how to build an agent capable of reading files, editing code, and executing commands.

A set of reliable PyTorch implementations of reinforcement learning algorithms designed to provide stable baselines for research and industry.

A tool that allows developers to copy UI elements from a browser and provide AI agents with the exact source code location and component stack.

A tool for creating LLM fine-tuning datasets by converting unstructured documents into high-quality structured QA pairs and evaluation sets.

A local-first issue tracker and task ledger designed for AI coding agents and humans to collaboratively track and operate work via CLI, TUI, and MCP.

A unified Python CLI and SSH workflow skill designed for AI agents to securely execute commands and manage files across remote servers on Windows, macOS, and Linux.

A suite of genomic language models designed to predict the functional effects of genome-wide variants using both aligned and unaligned genomic sequences.

A self-hosted AI accounting app that automates expense and income tracking by extracting data from receipts and invoices using LLMs.

An open-source PyTorch toolkit for Conversational AI that provides recipes and tools for speech and text processing, including speech recognition and synthesis.

A sovereign AI agents platform that allows users to build, deploy, and manage isolated autonomous agents on their own infrastructure with full observability and governance.

A Ruby-native AI framework that provides a consistent API for building chats, agents, and multimodal features across multiple AI providers in Ruby and Rails applications.

A lightweight reinforcement learning framework for training motion controllers to imitate reference motions across multiple simulator backends.

A local review tool that provides a visual interface for commenting on AI-generated plans, code diffs, and live web pages to provide precise feedback to AI agents.

An open-source collaborative OS for AI agents that provides a unified workspace where multiple agents can share threads, files, and a browser to work together.

An AI-powered search engine that uses a generative UI to render rich, cited answers with interactive components instead of plain text.

A CLI security scanner that discovers AI agents (Claude, Cursor, etc.), their MCP servers, and skills on your machine, and checks them for prompt injection, malicious code, and other vulnerabilities.

A native macOS terminal with a programmable control API designed to help developers manage and monitor multiple concurrent AI coding agent sessions.

A tool output proxy that intercepts command-line calls from AI agents to provide fabricated or modified responses for red-team testing and guardrail bypass evaluation.

An agent-first AI video production pipeline that replaces one-shot prompting with a governed workflow of planning, approval gates, and verified rendering.

A multi-agent framework for AI-assisted novel writing that coordinates planning, drafting, and style distillation to produce high-quality, coherent fiction.

An AI agent framework for ESP32-series IoT devices that enables users to define device behavior through chat and execute decision-making locally on the edge.

An open-source benchmark and execution harness for evaluating the ability of LLM agents to perform realistic legal work across various practice areas.

An Obsidian vault template that provides AI coding agents with persistent, cross-session memory through automated hooks and semantic search.

A Zotero plugin that automates academic paper reading by using LLMs to generate Markdown notes, mind maps, and literature reviews.

A local CLI tool that indexes and searches past coding agent sessions to provide agents and developers with exact recall of historical decisions and tool calls.

AReaL is an open‑source, fully asynchronous reinforcement‑learning framework for training large reasoning and agentic models. It offers modular micro‑services, supports many RL algorithms, runs on GPUs, NPUs or Ray clusters, and ships with dozens of ready‑to‑run examples (math, coding, search, vision‑language, RLHF, etc.).

An ultra-low-latency, self-hosted speech-to-text pipeline that enables real-time transcription and translation using advanced simultaneous speech research and multiple model backends.