Trending AI projects

An open-source, local-first coding agent for the terminal that lets developers use their own AI models to write and refactor code while keeping data private.

A collection of skills and plugins that provide AI coding agents with specialized knowledge and guidance for building and managing resources on the Databricks platform.

An open-source template for deploying a personal AI agent with persistent memory and integrations for Slack, GitHub, and Linear.

A minimal chatbot template built with Next.js and the Vercel AI SDK, featuring streaming chat, tool calling, and a polished shadcn/ui interface.

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

A MuJoCo-powered simulation framework and benchmark suite for robot learning, providing modular tools to create environments and test AI control algorithms.

A Python library for building swarm-style multi-agent systems using LangGraph, enabling specialized agents to dynamically hand off control and context to one another.

A travel search skill for AI agents that connects them to Fliggy's real-time inventory for flights, hotels, and attractions via a CLI.

A high-fidelity parametric 3D human body model that uses identity, pose, and facial expression parameters to generate realistic, differentiable human meshes.

A modular Swift SDK for audio processing on Apple Silicon, providing a unified API for TTS, STT, VAD, and audio codecs using the MLX framework.

A local memory operating system for Hermes Agent that provides permanent, multi-layered memory and context injection to prevent agents from forgetting across sessions.

PapersGPT is a native C++ Zotero plug‑in that indexes thousands of PDFs locally and lets you query them with offline LLMs. It offers millisecond‑fast retrieval, cross‑paper synthesis with clickable citations, and an AutoPilot mode for batch processing, all without sending data to the cloud.

brooks‑lint is an AI‑powered code‑review skill that evaluates code and tests against six decay‑risk dimensions derived from twelve classic software‑engineering books. It produces structured findings (Symptom → Source → Consequence → Remedy) with book citations, severity labels, and a 0‑100 health score, and can generate Mermaid architecture graphs, debt roadmaps, and an auto‑fix “sweep” mode. Distributed as Agent‑Skills, it installs on many AI platforms (Claude, Gemini, Codex, Cursor, etc.) via a single script and is configurable via a .brooks‑lint.yaml file.

claude‑tap is a Python‑based local proxy and web UI that records every request/response of AI coding agents (Claude, Codex, Gemini, Kimi, etc.). It lets you view prompts, tool calls, streaming output, token usage, and diff adjacent turns, all on‑machine with optional export to a self‑contained HTML file.

OpenSpline is a developer toolkit that adds synchronized, talking-head video avatars to existing voice AI agents, supporting both cloud and self-hosted GPU deployments.

Pandera is a data validation framework that allows scientists and engineers to define schemas for dataframes, ensuring data quality and robustness in processing pipelines.

A Python library for programmatically generating and editing CapCut project drafts, enabling automation of video editing tasks like media placement, text styling, and effect application.

A one-click installation kit for running Qwen3.8-27B on consumer NVIDIA GPUs, featuring automated VRAM profiling and an OpenAI-compatible API.

A collection of automation scripts that enables AI agents to build, test, and interact with iOS apps in the simulator using semantic accessibility APIs instead of fragile pixel-based coordinates.

ANOLISA is an operating system layer for AI agents that optimizes token usage, provides secure execution environments, and enables agent interaction directly within the terminal.

A local-first audio and video transcription tool that converts recordings into structured Markdown notes and enables AI-powered Q&A over the content.

PennyWise AI is an open‑source Android app that reads bank transaction SMS on your phone, parses them locally and stores a private expense ledger. It offers on‑device AI chat (Qwen 2.5), smart auto‑categorisation rules, multi‑currency analytics, budgeting tools, and export/backup features, all without sending any financial data off‑device.

A privacy-first, open-source speech-to-text application that uses NVIDIA's Parakeet model to provide fast, local transcription for 25 European languages.

A fully open‑source causal transformer trained from scratch on Brazil’s entire public legislative archive (≈360 GiB, 930 k documents). Includes a custom byte‑level tokenizer, a pure‑PyTorch 6‑layer decoder, reproducible data pipelines (R for ingestion, Swift OCR, graph‑based deduplication), and publicly released model and datasets on Hugging Face. Designed for research on Portuguese legal text, with rigorous split isolation and cryptographic auditability.

An industrial-grade AI short drama production pipeline for Claude Code that transforms ideas into complete video series using a structured 5-step workflow and dreamina CLI integration.

A rule-enforcement tool for AI coding agents that catches and forces repairs for project-specific rule violations that standard linters cannot detect.

A set of AI agent skills that enable end-to-end searching, comparing, and booking of worldwide hotels and flights through natural language conversations.

A plugin marketplace for Claude Code that provides AI-orchestrated application security tools to find, validate, and fix vulnerabilities in code.

A suite of 152 Claude skills that automate the generation and review of structured Excel deliverables for automotive safety, cybersecurity, and quality engineering standards.

A powerful search engine plugin for Obsidian that uses the BM25 algorithm to provide instant, relevance-ranked search across notes, PDFs, and images.

LinuxCNC is an open-source software suite for controlling CNC machines, including milling machines, lathes, 3D printers, and robot arms.

A Home Assistant integration that transforms any web browser or Android tablet into a private, hands-free voice assistant satellite with on-device wake word detection.

WARP (Weight‑Aware Runtime and Paging) is a C‑only inference engine that streams mixture‑of‑experts model weights from fast storage, using a bounded RAM cache. It enables running trillion‑parameter models like Kimi K3 on consumer laptops (e.g., 0.6 tokens/s on a 64 GB MacBook) and supports multimodal text‑image generation. No external libraries or GPUs are required; you build it with `make` and run models via the `waste` CLI.

AIDE ML is an open‑source Python package that implements an LLM‑guided tree‑search agent for automatically writing, debugging, and benchmarking machine‑learning code. Users specify a dataset, a natural‑language goal, and an evaluation metric; the agent iteratively generates and refines Python scripts, pruning the search tree based on metric feedback. The repo includes a CLI, a Streamlit UI, HTML visualisation, Docker and local‑LLM support, and is designed for both research on agent architectures and practical ML pipeline generation.

PhoneCLI is a framework that compiles mobile app navigation into deterministic commands to reduce VLM dependency, lowering costs and latency for mobile AI agents.

An intelligent CLI tool that analyzes your hardware to recommend, rank, and manage local LLMs, featuring hardware-calibrated memory estimation and an integrated MCP server for AI assistants.

A VST3/AudioUnit plug-in that integrates the Neural Amp Modeler engine into DAWs for AI-driven guitar amplifier modeling.

Live Avatar is a research‑grade system that turns spoken audio (plus an optional reference image or text prompt) into a continuous, real‑time video of a talking avatar. Built on a 14 B diffusion backbone (WanS2V‑14B) with a LoRA fine‑tune, the authors achieve ~45 FPS streaming on multi‑GPU H800 hardware and can generate videos longer than 10 000 seconds via block‑wise autoregressive processing. The repo provides Conda‑based installation, multi‑GPU and single‑GPU inference scripts, optional FP8 quantisation and JIT compilation for speed, and a Gradio UI. Model weights are hosted on Hugging Face under Apache 2.0.

A high-performance implementation of Stable Diffusion optimized for CPUs, utilizing OpenVINO and Latent Consistency Models to enable near real-time image generation.

A terminal-native, AI-powered image search tool that lets you search local folders using natural language or example images without needing to upload files.

A real-time security layer for AI agents that blocks destructive commands, scans plugins for malicious code, and performs daily security audits to prevent data theft and system damage.

An MCP server that reduces AI coding agent token usage and costs by providing structural code navigation, persistent memory, and Bash command rewriting.

okf‑skills is a Claude Code‑native toolkit that lets LLM agents create, validate, visualize and back‑fill knowledge bundles in the Open Knowledge Format (OKF). It ships as a Claude plugin, a set of `skills.sh` agent skills, a GitHub Action, and a read‑only MCP server, handling trust, provenance and staleness metadata defined in OKF v0.2.

An open-source framework for embodied foundation models that enables robots to perform complex physical manipulations by bridging vision-language understanding with action prediction.

kraken is an open‑source, Python‑based OCR toolkit for historical and non‑Latin scripts. It provides trainable layout analysis, reading‑order detection, and neural‑network character recognition, supporting right‑to‑left, bidirectional and vertical scripts. The tool outputs standard XML formats (ALTO, PageXML, hOCR), offers word‑level bounding boxes, and ships with a public model repository. Install via pip/pipx on Linux or macOS, fetch a model with `kraken get …`, and run a one‑line command to binarise, segment, and OCR an image.

A Swift package that provides a unified, drop-in replacement for Apple's Foundation Models framework, enabling support for multiple on-device and cloud LLM providers.

A toolset for creating professional 10-page visual carousels from AI concepts or papers, featuring automated layout rendering and strict factual verification.

OSTT is an open‑source, terminal‑native voice‑to‑text CLI for Linux/macOS. It records via a global hotkey, transcribes with local Whisper models or many cloud providers, and outputs to clipboard, files, stdout, or custom shell commands. Supports GPU acceleration, per‑model parameters, history/retry, and scriptable post‑processing actions.