Trending AI projects

ChattyPlay-Agent is a multi-functional web platform that integrates AI services, media parsing, productivity tools, and financial data into a single, unified interface.

A Python library from Neo4j that lets you build Graph Retrieval‑Augmented Generation (GraphRAG) apps. It provides pipelines to turn text/PDFs into a Neo4j knowledge graph, utilities for vector indexing, hybrid retrievers that blend vector search with graph traversal, and a simple RAG API that plugs any supported LLM (OpenAI, Anthropic, Ollama, etc.). Install via `pip install neo4j-graphrag` (optional extras for your LLM), create a KG, index embeddings, then query with natural language.

A real-time, browser-based motion capture tool that uses a webcam to drive MMD characters and export animations as .vmd files.

An AI agent framework for Claude Code that guides developers through a structured seven-wave process from idea to code, enforcing TDD and human review at every step.

DEIMv2 is a family of real-time object detection models that builds on the DEIM framework and leverages DINOv3 features. It offers a range of sizes from ultra-light to extra-large, achieving state-of-the-art accuracy on COCO (S model >50 AP) while supporting easy fine-tuning and deployment.

A library for training and analyzing sparse autoencoders to improve the mechanistic interpretability and safety of neural networks.

SoulX-FlashTalk is a system for real-time, infinite streaming of audio-driven avatars using self-correcting bidirectional distillation to create continuous talking head animations.

A lightweight, GPU-independent OCR tool for digitizing books and magazines, optimized for standard home computers and supporting both printed and handwritten Japanese text.

An AI-native interactive shell for Linux that allows users to execute system operations via natural language while providing risk-based security controls and full PTY support.

ResearchPilot‑Skills is an MIT‑licensed set of Claude/Code‑buddy “skills” that automate the full lifecycle of an AI research project—literature review, experiment design, code implementation, iterative debugging, and LaTeX paper writing (including bilingual translation). Install scripts drop 17 skill directories into the user’s Claude skill folder, then you drive the workflow with simple slash commands (e.g., `/research[A]-exploration`). The tool enforces documentation before code changes, generates journal‑quality figures, and maintains versioned logs of design, code, and manuscript drafts.

OpenUnlearning is an extensible Python framework that unifies LLM unlearning benchmarks (TOFU, MUSE, WMDP, DUET), implements 12+ forgetting methods, supports 10+ evaluation metrics, and works with many LLM families. It uses Hydra configs, provides Docker images, scripts for baseline runs, and a community leaderboard, making it easy to run, compare, and extend unlearning experiments.

Performance Monitor is a free, open-source alternative to expensive SQL Server monitoring tools. It collects performance data without phoning home, offers real-time alerts, and can hand alerts to your own AI agent.

A collection of modular extensions for the Pi terminal-based coding agent that add features like session management, improved permission handling, and custom interface tools.

A living benchmark for symbolic regression that provides a unified, reproducible framework to compare modern SR methods using diverse datasets and scikit-learn compatible APIs.

An open-source marketplace repository for AI agent skills, providing a secure distribution system and CLI for adding capabilities to tools like Claude Code and Codex.

An open-source agent platform for project work and multi-agent collaboration that can be deployed as a desktop app, web server, or embedded Python library.

An automated installer and manager for the Pi coding agent that provides a curated, pre-configured setup of tools and packages to eliminate manual configuration.

A Chrome extension that uses local image embeddings to find and remove duplicate photos from a Google Photos library without requiring OAuth setup.

A comprehensive guide on best practices for creating structured agent skills to improve AI agent reliability, discoverability, and token efficiency.

BioMCP is a CLI tool and MCP server that lets users query 72 biomedical databases (PubMed, ClinVar, ClinicalTrials.gov, etc.) with a single, uniform command grammar. It supports unified searches, cross‑entity pivots, enrichment, batch retrieval, and local study analytics, and can be integrated into AI agents like Claude Code or Codex via its MCP interface.

The repository …

Plane MCP Server is a Python service that exposes Plane’s project‑management API as Model Context Protocol tools, letting LLM agents read and modify Plane resources via stdio or HTTP (OAuth or API‑key). It offers 30 resource‑specific tools, PQL querying, and backward‑compatible naming, all under an MIT license.

A collection of opinionated configurations, sandboxing rules, and hooks for Claude Code to enable secure, high-throughput autonomous AI coding.

A JavaScript library for client-side indecent content detection using TensorFlow.js to classify images into safe and NSFW categories.

BrowserTools MCP is an MCP server paired with a Chrome extension that lets AI coding agents observe the developer's actual browser session - console output, network activity, screenshots, and Lighthouse audits - without spinning up a separate automated browser. It keeps the agent in sync with the page you're already on, including your logins and state.

An MCP server that enables AI agents to query, explore, and analyze data within ClickHouse clusters or local chDB instances using natural language.

An MCP server for AI coding assistants that enables automated JavaScript reverse engineering using the Camoufox anti-fingerprint browser.

PySINDy is a Python package that implements the Sparse Identification of Nonlinear Dynamical systems (SINDy) algorithm, letting users discover explicit governing equations from time‑series data. It offers multiple sparse optimizers, customizable feature libraries, and produces interpretable models useful for prediction, control, and scientific insight.

A workflow pack for AI coding agents that provides structured processes for formal verification and theorem proving in Lean 4.

A tool for creating short vertical story videos by animating the transition from hand-drawn line art to colored images.

A real-time dense feed-forward scene reconstruction system for 3D mapping and localization, featuring optional open-set 3D object detection.

SciWrite is an AI-powered editorial skill that reviews scientific and engineering manuscripts for clarity, structure, and style based on established academic writing principles.

Comfy‑Org/docs is the source repository for the multilingual documentation site of ComfyUI, an AI‑focused node‑based UI. It uses Mintlify to build the site, offers local preview, automated translation (via OpenAI‑compatible models), API reference generation from OpenAPI specs, and safeguards (redirects, guarded files) to keep docs reliable. Contributors edit MDX, open PRs, and Vercel deploys the live docs.

A lightweight cli for running single-purpose AI agents. Define focused agents in TOML, trigger them from anywhere; pipes, git hooks, cron, or the terminal.

A Python framework for audio processing that brings Digital Audio Workstation (DAW) features like VST plugin support and Faust signal processing to Python scripts.

A Kubernetes operator that automates the lifecycle of Flux CD, providing self-service ephemeral environments and AI-assisted GitOps management.

A hat-based orchestration framework that keeps AI agents in an iterative loop with automated validation gates until a task is fully completed.

A terminal and JSON CLI tool, written in Rust, that indexes the local conversation history of dozens of different AI coding agents into a single searchable database. It provides high-speed full-text search, optional local semantic search, and a 'robot mode' API so AI agents themselves can query past sessions. It emphasizes privacy (everything local), robustness (atomic index updates, quarantine of corrupt data), and performance (sub-60ms searches).

GenAI Prices is a multi‑language library that aggregates and normalises LLM inference pricing from dozens of providers, offering APIs/CLIs to calculate token‑level costs and detect price discrepancies.

A research codebase that implements a dynamic, graph‑based memory system (Agentic Memory) for LLM agents, reproducing the experiments from the NeurIPS 2025 paper.

A curated collection of research papers and resources on Composed Image Retrieval (CIR), covering attribute-based, supervised, few-shot, zero-shot, and other approaches, with evaluations across multiple benchmark datasets. It accompanies a comprehensive survey and is continuously updated.

CyberBattleSim is a research platform that uses reinforcement learning to simulate interactions between automated attackers and defenders in abstract enterprise network environments.

A curated directory of resources, research papers, and tools focused on offensive AI, adversarial machine learning attacks, and corresponding defensive strategies.

AI Weekend Builds is a collection of 20 AI projects across four volumes, each with starter code, a full README, and copy-ready prompts. Projects range from easy (1–3 hours) to advanced (a full day) and cover domains like personal data, local-first life tools, UI-to-code, and autonomous agents. Volume 4 uses an OpenAI API key; earlier volumes use Anthropic. The repo is aimed at developers comfortable with Python or Node.js and terminal commands.

This repository is the official code companion for the O'Reilly book *Transformers - The Definitive Guide*. It provides organized Jupyter notebooks for every chapter, covering a wide range of transformer applications from time series and vision to agents and deployment, all runnable directly in Google Colab.

A project for Apple Silicon that optimizes vision-language model inference by using shared-context batched scoring instead of autoregressive text generation to answer multiple questions about an image.

A web API service for solving CAPTCHAs using model inference, supporting local and public network deployment.

BugTraceAI is an open‑source, self‑hosted suite for AI‑augmented web and API pentesting. It consists of four interoperable components—CLI (terminal UI + REST/MCP engine), WEB dashboard, API‑only service, and a Launcher installer—plus model checkpoints on HuggingFace and a vulnerable demo app (BugStore). The AI agents plan and prioritize scans, call deterministic security tools, and produce evidence‑rich reports. Installable via a one‑liner script, it supports multiple LLM providers (OpenRouter, Anthropic, Z.ai, Bedrock, Ollama) and is Apache‑2.0 licensed.