Discover 1000's of New AI Tools and Projects

Budge is an open-source agent skill for iterative UI tuning. It lets developers nudge values up and down, switch between properties, and quickly copy prompts, making design and layout refinement faster inside tools like Claude Code, Codex, and Cursor.

LiteParse is a model-free, open-source document parser built for AI agents and production document processing pipelines. It focuses on fast parsing across many file formats with zero cloud dependency, and is positioned as a core building block for agentic document workflows.

Browser Harness is a browser automation harness designed for LLM-driven workflows. It interacts directly with Chrome through CDP, avoids heavyweight browser frameworks, and can self-heal by modifying helper code on the fly so agents can complete a wide range of browser tasks with fewer constraints.

SuperCmd is an open-source macOS productivity tool that replaces copy-paste AI workflows with system-wide inline prompting and utility actions. It combines AI-assisted writing and editing with command-style utilities, and recent updates highlight powerful built-in calculations, conversions, and workflow shortcuts.

TourKit is a library for adding polished, Apple-style onboarding experiences to Mac apps. It is aimed at developers who want a guided first-run experience with native-feeling presentation and minimal implementation overhead.

Agentic Inbox is an open-source email inbox you can host yourself, with a built-in AI agent running entirely on Cloudflare Workers. It is designed for people who want programmable email workflows, local control over deployment, and agent-powered inbox operations without relying on a closed hosted system.

Dinky is a lightweight image optimization app created as a fast replacement for Optimage. It is positioned as a compact utility, around 4.3 MB, for users who want a simpler, faster desktop image optimization experience.

EasyInvoicePDF is a free and open-source invoice generator built with Next.js and hosted on Vercel. It is aimed at generating clean PDF invoices quickly, with a simple developer-friendly stack and straightforward deployment model.

SuperIsland brings a Dynamic Island-style experience to the Mac and turns the notch into a programmable widget space. It includes media controls, utilities, and extensions, with recent updates adding features like a teleprompter and display placement controls.

Multica Autopilot is an open-source system for running Claude Code-style routines locally with multiple agent harnesses, including Opencode, Codex, Hermes, and OpenClaw. It is designed to make routine execution portable across local agent setups.

Fire-PDF is a Rust-based PDF parsing engine focused on speed and structured extraction for LLM and agent workflows. It aims to convert PDFs to markdown faster, preserve formulas, and extract full tables with minimal configuration.

Omi for Desktop is an open-source local desktop assistant positioned as a life architect. It can observe your screen, listen to conversations, and help guide what to do next, aiming to function like a second brain that actively pays attention.

OpenMed is an open-source medical AI toolkit that runs locally on Apple Silicon and iPhone with no cloud or API dependency. It includes an MLX backend, Swift package support for macOS and iOS, and a large collection of PII detection models across multiple languages.

OpenClaude is an open-source coding agent CLI that mirrors Claude-style workflows and keeps shipping new capabilities. Recent releases add features like a loop command, monitoring tools, Docker GHCR support, and ongoing quality improvements for local agent-driven development.

T3 Code is a minimal web interface for coding agents, currently focused on tools like Codex and Claude. It is built to work across local, web, and remote environments, making project creation and agent-driven coding workflows easier to manage from a lightweight UI.

Open Agents is an open-source coding agent that runs in the cloud and has been used to write production code, including parts of itself. It is aimed at developers who want a transparent, hackable alternative for cloud-based agent-driven software development.

DiffKit is a fast, design-first GitHub dashboard for developers who want to stay on top of pull requests, issues, and code reviews without the usual noise. It focuses on clarity, speed, and a cleaner workflow for code collaboration.

Wikiwise is an open-source Mac app for building and managing your own Karpathy-style LLM wiki. It uses markdown and HTML under the hood, makes local setup simple, and supports one-click public sharing for people building personal knowledge bases with agents.

kimün is a terminal-based note-taking app built in Rust with ratatui. It uses local Markdown files and offers fast indexing, powerful search, backlinks, wikilinks, and Neovim-style interaction for people who want a keyboard-first knowledge tool.

Matter CLI lets agents interact with a user’s Matter account to summarize inbox content, find articles, organize tags, and retrieve highlights. It is designed as a bridge between reading workflows and agent automation across tools like Claude Code and Codex.

Orca IDE is a free and open-source IDE for agentic engineering across Windows, macOS, and Linux. It supports multiple agent harnesses, including Claude Code, Codex, Pi, and KimiCode, with built-in worktree management for parallel development workflows.

ParseBench is an open-source document OCR and parsing benchmark designed for agent knowledge work. It measures parsing quality for real-world files and is intended to help teams evaluate document understanding systems for AI agents and production workflows.

pi-executor gives agents a shared execution layer for MCP, OpenAPI, and GraphQL APIs through a single discovery and calling interface. It is built to keep large API schemas out of context, centralize auth handling, and simplify interactive tool use for agents.

Metabase 60 is a new release of the open-source analytics platform Metabase, positioned around expanded AI capabilities. It continues Metabase's focus on accessible business intelligence while making AI-powered workflows and data interaction more available inside an open-source product.

Paperclip Desktop packages the Paperclip experience into an easy-to-use Mac app. It is aimed at users who want a simpler desktop wrapper around Paperclip functionality, with free access and straightforward installation for macOS users.

isolet lets developers package components into self-contained widgets that can be embedded anywhere without host-page styles breaking them. It works across React, Solid, Svelte, and vanilla JavaScript, making reusable embeddable UI components easier to ship.

Superconductor is a native macOS app for agentic engineering built in Rust, without Electron or Tauri. It is designed to help developers manage coding agents with a fast local UI and low-friction workflows.

YOLO26-MLX is a native Apple Silicon implementation of YOLO26 built without PyTorch or external GPUs. It is positioned as a fast open-source release for local training and inference on Apple hardware, with performance gains for both workflows.

Print pay slip is a small payroll-related web app built with Claude Code. Based on the shared post and live demo reference, it appears focused on generating printable pay slips with a simple browser-based workflow.

The /how skill is an open-source agent skill designed to help understand how a system works, not just patch symptoms. It is meant to improve root-cause analysis, architecture comprehension, and refactor planning for both humans and coding agents.

Sparkle v4 is a desktop organization agent from Every. It examines a user's filesystem, helps design a folder structure that fits their way of working, and begins with cleanup tasks like removing old installers and duplicates.

Superblocks 2.0 helps teams build AI-generated enterprise apps while keeping IT visibility, permissions, and governance in place. It is positioned as a safer enterprise alternative to uncontrolled vibe-coded internal apps running on production data.

This project is an upcoming markdown editor focused on extremely fast cold starts, rapid file reading and indexing, and a lightweight writing experience. It is presented as a free and open-source tool aiming for speed comparable to top native editors.

This tool is presented as a product that turns a described hardware idea into a prototype design. It appears positioned as an idea-to-prototype workflow for hardware concepts through a simple web-based experience.

Copilot CLI Remote Control lets users run Copilot on a local machine and control the agent remotely from a phone. It is positioned around mobile access to a home or office coding machine while keeping the main environment plugged in and available remotely.

infinite is a spatial terminal workspace designed for multitasking across many grouped terminal contexts. Instead of relying only on tiling or pane splits, it lets users pan across a larger workspace to organize parallel threads of work.

ParclLabs AI Connect is shown powering targeted property intelligence workflows, such as finding homes that match specific criteria, clustering them geographically, ranking neighborhoods, and dispatching marketing actions. It appears focused on AI-assisted real estate and spatial targeting workflows.

Vost helps developers take a Claude Managed Agent and launch, package, and monetize it quickly. It is positioned as a fast path from agent creation to distribution and commercialization.

This post highlights LiteParse as part of the LlamaIndex ecosystem, emphasizing fast parsing, broad format support, zero cloud dependency, and use in agent workflows. It is effectively a LlamaIndex ecosystem entry point for the LiteParse project.

pi-steel connects the Pi coding agent to a Steel browser, giving the agent browser capabilities through a simple install command. It is presented as an integration for agent-driven browsing and web interaction.

agentchat is an open-source skill that teaches agents to participate in shared group chats together across machines. It is aimed at agent communication workflows, letting multiple agents exchange context and coordinate in chat-based environments.

This post presents OpenMed 1.0.0 as a local medical AI toolkit that runs without cloud APIs on Apple hardware. It highlights MLX support, Swift packaging, and privacy-friendly local execution across macOS and iOS.

This entry captures pi-executor as a shared execution layer that keeps large schemas out of context and gives agents one place to discover and call APIs. It is designed to simplify auth, tool discovery, and interactive flows for agent systems.

This product extends the Frontend Slides concept into AnyGen, letting users create editable and shareable HTML slides through a web interface without needing a coding agent. It is positioned as an easier path to polished browser-based presentation building.

Supabase Agent Skills are official instructions and workflows that teach agents how to work with Supabase using best practices. They cover security, RLS, docs, schema management, and CLI or MCP usage for safer agent-driven development.

This entry captures a direct product release shared via a short download link post. The public post provides limited context, so this listing represents the release reference itself rather than a fully expanded project profile.

Locker is presented as an open-source alternative to Google Drive, Box, and Dropbox. It is provider-agnostic, supports S3 or local storage backends, and includes a virtual file system layer.

Better Agent is a composable agent framework for TypeScript. It is designed for developers who want to assemble and customize agent systems with a modular approach rather than relying on a fixed agent abstraction.

Cabinet is presented as an open-source version of Claude Managed Agents. It is positioned for users who want a transparent, self-directed alternative for managed-agent style workflows.

Sendblue CLI is now open source, making it easier for agents and developers to inspect the codebase and contribute improvements. It is positioned as a command-line interface for Sendblue-related messaging workflows.

This entry highlights a T3 Code workflow improvement that makes adding new projects easier without depending on the native file browser. It focuses on smoother setup in web and remote environments for coding-agent workflows.

This Omi entry emphasizes the product as a life architect, a local assistant that can observe screen activity, hear conversations, and suggest what to do next. It is positioned as a second brain that pays attention continuously.

This entry captures the OpenClaude 0.3.0 release, highlighting additions such as a loop command, monitoring support, Docker GHCR support, and multiple bug fixes for the open-source coding-agent CLI.

This project is presented simply as a macOS terminal for AI agents. It appears to focus on a local terminal experience tailored to agent workflows on Mac.

insforge is presented as backend infrastructure for projects where AI agents can build much of the product but struggle with auth, databases, storage, and backend setup. It is positioned as a way to remove backend friction from vibe-coded applications.

skye is described as an agentic home screen for iPhone, rethinking the static icon grid with a more context-aware and adaptive mobile interface. It is positioned as a more intelligent front layer for the phone.

This tool gives users multiple expert Claude Codes focused on different areas of a codebase, such as auth, database, and API. It supports parallel work, sub-swarms, and shared memory or task structures for multi-agent development workflows.

This tool is presented as a simple site scoring system for context, reverse engineered and shared after shipping improvements to VV. It appears to evaluate how well a site conveys usable context for downstream understanding.

This project is a family trip operations dashboard that includes route playback, convoy tracking, meal logistics, family readiness, expense tracking, and a shared operational timeline. It is a playful but functional planning tool for group travel coordination.

DFlash brings speculative decoding to Apple Silicon using stock MLX without a fork. It is positioned as a performance-focused inference enhancement for local model execution on Apple hardware.

This entry captures the core LiteParse launch framing from Jerry Liu, positioning it as a model-free, open-source document parser for AI agents and a central building block in agentic document processing.

This entry captures the original Browser Harness launch framing, emphasizing direct Chrome DevTools Protocol access, self-healing helper edits, and fewer framework constraints for browser tasks driven by AI agents.

This entry captures the original source post for Agentic Inbox, focusing on the open-source, self-hostable email inbox with a built-in AI agent running on Cloudflare Workers.

This entry captures Metabase's framing around an AI-focused open-source milestone in the Metabase 60 release cycle. It emphasizes the product's open-source analytics identity alongside new AI capability messaging.

This entry captures the original launch post for EasyInvoicePDF, presenting it as a free and open-source invoice generator built with Next.js and hosted on Vercel.

This entry captures the original Multica Autopilot post, centered on running routines locally with multiple agent harnesses such as Opencode, Codex, Hermes, and OpenClaw.

This entry captures the original Fire-PDF launch framing, emphasizing speed, markdown conversion, formula preservation, and table extraction for LLM and agent document workflows.

This entry captures the original Matter CLI announcement, which frames the tool as a way for agents to access a user's Matter inbox, highlights, tags, and reading workflows directly.

This entry captures the original Orca IDE announcement, positioning it as a free and open-source IDE with support for multiple agent harnesses and built-in worktree management.

This entry captures the original launch framing for Open Agents, a cloud-based coding agent that the creator says has written every shipped line of code, including parts of itself. It emphasizes open sourcing and transparent agent-driven development.

This entry captures LM Studio becoming an official OpenClaw provider, enabling users to run local models privately and for free on Mac, Windows, and Linux through the OpenClaw stack.

This entry captures the original ParseBench announcement, emphasizing document OCR benchmarking for agent knowledge work and evaluation of parsing quality on real-world files.

This entry captures a SuperCmd workflow focused on a Cursor-style command palette for the entire Mac, letting users trigger inline AI prompts on selected text and replace copy-paste writing workflows.

This project is a curated list of official or maintained agent skills from real teams and companies. The post highlights a new lightweight UI layered over a GitHub-maintained collection to make discovery easier.

This project turns the Chrome new tab page into a custom dashboard for managing too many tabs, grouping them by domain, surfacing easy wins, and adding more playful browser interaction patterns.

This entry captures the concept of personal open source software, centered on software created and tailored around an individual's own workflows, needs, and AI setup rather than packaged one-size-fits-all products.

This entry captures the earlier launch framing for Paperclip Desktop, a Mac app that packages the Paperclip experience into an easier-to-use desktop form.

This entry captures Poke as an AI product on iMessage, characterized in the post as OpenClaw for mainstream users and notable for scaling successfully on iMessage.

This entry captures the original Wikiwise launch framing, focused on a Mac app that scaffolds and manages a personal Karpathy-style LLM wiki using markdown and HTML under the hood.

This entry captures a direct release post for Glass. The post primarily shares a download link, so the listing is based on the release reference rather than a fully expanded project page.

Mole is a Mac cleaning tool designed to reclaim significant disk space quickly. The release post highlights update changes, large user adoption, and automation around cleanup and optimization tasks.

This entry captures the v2.0.0 release of Paperclip Desktop, a Mac app that packages the Paperclip experience into an easier-to-use desktop application for macOS users.

This entry captures the Hermes agent self-evolution project from Nous Research, framed around prompt self-improvement and evolutionary optimization for agent behavior.

This entry captures a small product or demo centered on mouse click interactions. The post itself gives little context beyond the interaction preview, so this listing is intentionally lightweight.

This entry captures a one-click local runner for open-sourced Claude Managed Agents, aimed at making local setup easier for developers who want to run these agent workflows on their own machines.

This entry captures a tool that generates a DESIGN.md-style artifact from a website. It is positioned as a way to extract design-system structure and prompts that AI agents can understand and reuse.

This entry captures an Obsidian Web Clipper update that adds interactive transcripts, extending the clipping workflow for richer media and content capture.

htop-style real-time monitor for AI coding agent sessions, tokens, context windows, rate limits, and ports.

AI-powered platform that uses agentic workflows to automatically generate product requirement documents (PRDs).

Semantic code search MCP plugin for Claude Code and other coding agents that turns large codebases into searchable context.

This entry captures an early AI transcription app for macOS that was built quickly with agents and is intended to be open sourced when it is ready.

This entry captures a Chinese-language launch post for SuperConductor, describing it as a native Rust macOS app designed for managing coding agents with strong performance and a smooth UI.

This entry captures a broader product and ecosystem framing around public open-source AI releases and real personal setups, emphasizing momentum in tools like GBrain and similar systems.

CoreML-LLM is presented here in a v0.4.0 release that brings Gemma 4 multimodal support to iPhone ANE, expanding local multimodal inference on Apple mobile hardware.

This entry captures a social proof style post praising Instant as a fast product for starting projects, now including free auth support and quick setup. It serves as a product mention and ecosystem pointer.

This entry captures the original launch post for SuperIsland, presenting it as a programmable Dynamic Island-style widget space for the Mac notch.

This entry captures the original YOLO26-MLX release framing, highlighting native Apple Silicon support without PyTorch or external GPUs and performance improvements for local training and inference.

Trading Arena is positioned as a simple way to deploy an AI trading agent by writing a strategy in plain English, choosing a model, setting risk limits, and tracking performance on a leaderboard.

FloatyDo is a minimal todo app focused on craft over feature bloat. It is positioned for users who like the feel of lightweight daily task lists and want something between analog scratch notes and polished task apps.

This tool lets users swap hub for inspect on a GitHub URL to chat with any repository locally in the browser using a Pi agent and a virtual file system on top of the GitHub API.

PayKit is positioned as the first billing framework for TypeScript, aiming to give developers a structured way to build payment and billing workflows into TypeScript applications.

opensrc is a developer tool that gives AI agents deeper implementation context by letting them access the actual source code of dependencies. This release highlights a Rust-based version.

This entry captures a vibecoded website animation workflow, combining multiple tools to describe, generate, animate, and ship playful interactive behavior on a website.

This entry captures a product or demo framed around the question of whether something is ready to ship. The post offers limited detail, so the listing stays intentionally lightweight.

This entry captures a hardware and home lab setup built to run OpenClaw, including a compact rack and local compute environment for self-hosted agent workflows.

This entry captures the original source post for SuperCmd as an open-source alternative to Raycast Pro, WisprFlow, and Speechify, centered on AI-assisted macOS productivity.

lunel is presented as a way to run coding tools like Codex, Opencode, Claude Code, and a full dev environment directly from a phone. It is positioned as a mobile remote-dev tool.

Algebrica is a mathematical knowledge base built over years, with hundreds of structured entries and no ads or distractions. It is positioned as a focused knowledge resource rather than a monetized content funnel.

Snippy is presented as a focus assistant that watches your screen and alerts you when you get distracted, aiming to help users stay on task during computer work.

Agent Editor is a visual interface for Mastra Studio that lets teams make agent changes without directly editing code, making collaborative configuration and iteration easier.

Helmor is an open-source agent orchestration IDE positioned as a fast, elegant, and low-friction environment for multi-agent development. The launch messaging frames it as a step beyond Conductor with simple migration.

Expect is an open-source tool for testing agent code in a real browser environment. It is designed to work with agent workflows while providing more realistic browser-level validation than mocked or purely synthetic tests.

Hosted and local MCP server that lets AI agents use Apify actors for scraping, crawling, search, and web automation.

This entry captures a prompt injection firewall for AI agents, framed as a small local tool for scanning tool outputs and reducing prompt injection risk across inputs like email, docs, and other external data.

Rowboat is an open-source, local-first AI coworker that compiles emails, meetings, and work activity into a living knowledge graph and uses it to help get things done. It is positioned as a voice-powered productivity and knowledge system.

Tegaki is a frontend library that turns text into animated handwriting without manual SVG path authoring. It supports React, Vue, Svelte, Solid, Astro, Web Components, and vanilla JavaScript, and includes bundled handwriting fonts plus tools for generating custom font bundles.

Moshi is a mobile terminal built for AI coding agent workflows. It lets you monitor and interact with Claude Code, Codex, and other terminal-based agents from your phone, with support for persistent mosh and tmux sessions, push notifications, and remote supervision of long-running coding tasks.

Spectrum is an Electron-based desktop app for managing coding projects with project dashboards, canvas-style workspaces, terminals, browser panels, file editing, and chat-oriented workflows in one place. It is built around the project layer surrounding coding and agent work, with lightweight memory and tracking backed by SQLite.

WikiKarp is an early-access knowledge tool positioned as an LLM wiki viewer. It appears aimed at making notes and knowledge more structured and usable in AI workflows, with the current public site offering a waitlist for early users.

ui.sh is a toolkit for coding agents like Claude Code, Cursor, OpenCode, Amp, and Codex that helps them build better user interfaces. It is presented as an invite-based product from the creators of Tailwind CSS and Refactoring UI.

Cogito is a Mac-native Markdown editor built for power users, developers, and agent-era workflows. It focuses on fast writing, native performance, plain-text files on disk, wiki links and embeds, rich preview support, document summaries, and integrations like Typefully publishing.

Hue is an open-source skill for Claude Code that turns a brand, URL, image, or short description into a complete design system in one prompt. It is positioned as a fast way to generate design language guidance for agent-assisted UI work.

MarkItDown is Microsoft's Python tool for converting PDFs, Word docs, Excel files, PowerPoints, images, audio, HTML, YouTube URLs, and more into structured Markdown for LLM and text-analysis workflows. It focuses on preserving useful document structure while producing token-efficient output for downstream AI systems.

Email Emulation is a skill for testing auth and email flows without sending real emails. It can reroute traffic through the Resend SDK, retrieve messages from a local inbox, and extract verification codes or magic links to complete flows in development and testing.

Agent Analytics Skill analyzes server logs to classify AI bots and crawlers, reveal what agents are asking about on your site, identify technical issues blocking discovery, and generate actionable recommendations for AI-driven traffic and growth.

Impeccable is an open-source design skill, CLI, and browser extension that teaches AI better design discipline. It adds design context, anti-pattern detection, critique tooling, and installable commands for harnesses like Cursor, Claude Code, Gemini CLI, and Codex CLI.

JetKVM is a hardware remote KVM solution for accessing and controlling a computer from anywhere without being constrained by the operating system. This post highlights the convenience of simple setup and OS-independent remote access.

This entry captures a local OpenClaw setup using Gemma 4 on a MacBook Air M4 with 16GB RAM, showing local model experimentation and lightweight agent workflows on Apple hardware.

Clicky is a Mac assistant that sits next to your cursor, watches what is on your screen, and lets you ask questions by voice while you work. It is framed as a real-time on-screen helper for creative tools, writing, design, editing, and setup workflows.

Boneyard is a framework for generating pixel-perfect skeleton loading screens directly from real UI layouts, avoiding manual measurement and hand-built placeholders. It supports React, Preact, Vue, Svelte, Angular, and React Native, with CLI and Vite-based capture workflows.

CocoIndex is an open-source indexing engine and data transformation framework for AI and search workloads. It is designed to help developers build indexing pipelines over changing data sources and power retrieval, semantic search, and agent-facing applications.

HiClaw is an open-source collaborative multi-agent runtime platform built around auditable rooms, manager-worker orchestration, and human visibility throughout agent work. It is designed for coordinated multi-agent execution in self-hosted and enterprise settings.

DeerFlow is an open-source super-agent harness for research, coding, and creation across long-running tasks. It combines sub-agents, memory, sandboxes, skills, and messaging-style workflows for more capable autonomous execution.

OpenClaude is a terminal-first open-source coding-agent CLI that works across OpenAI-compatible APIs, Gemini, GitHub Models, Codex, Ollama, and other providers. It keeps prompts, tools, agents, MCP, and streaming workflows in one interface.

Hermes Agent is an open-source persistent AI agent built around a learning loop, long-term memory, self-improving skills, messaging gateways, and flexible runtime backends. It is designed to run beyond a single laptop session and grow with continued use.

Open-source AI workflow automation platform and Zapier alternative with a large MCP-compatible integration ecosystem.

Open-source AI workflow automation platform with large MCP-compatible integration support and no-code builder.

Collection of specialized AI agents and workflows for tools like Claude Code, OpenClaw, Cursor, and others.

General-purpose AI agent framework with persistent memory, multi-agent delegation, terminal execution, and SKILL.md support.

Browser automation CLI for AI agents with Playwright-compatible controls and a fast native Rust core.

Reusable single-VM agent runtime with deterministic context assembly, task intake, persistence, and operator visibility.

Public collection of AI agent skills spanning analysis, skill building, OSS readiness, git workflows, and more.

AI-assisted Solana app builder that combines skills, MCPs, CLIs, launch, and fundraising workflows for crypto app creation.

AI-native workflow language and runtime for deterministic multi-step agent execution with MCP and agent integrations.

Large medical research skills library for AI agents covering evidence review, protocol design, data analysis, and academic writing.

Single-file Claude Code guideline pack derived from Andrej Karpathy’s observations on LLM coding pitfalls and execution quality.

Privacy-focused product spanning MCP server, API, CLI, browser extension, and encrypted/alias-style workflows.

Claude Code skill and slash command that finds quick-win OSS issues, runs a TDD contribution loop, and opens GitHub PRs end to end.

Large open-source collection of runnable LLM apps, agents, and RAG templates across many categories.

Offline CLI that analyzes Claude Code context usage, detects token waste, and generates actionable cleanup and optimization commands.

Security intelligence MCP server that gives AI agents CVE, EPSS, KEV, ATT&CK, Shodan, VirusTotal, and related vulnerability tooling.

Community directory of agent skills, tutorials, and related tooling with an associated marketplace presence.

Reusable agent skills collection covering analysis, OSS readiness, git workflows, and related development tasks.

Local-first memory orchestration system for AI agents with durable writes, multi-sink fanout, retrieval learning loops, and operator controls.

Open-source toolkit for evolving coding-agent skills from benchmark failures into stronger specialist behaviors.

Agent framework and app environment aimed at moving beyond toy demos into more operational AI workflows.

Base framework for building custom AI agents with pluggable tools, models, and workflows.

Agent skills repository including security-focused and repo-analysis workflows.

Open-source Claude Code operating system from Garry Tan with 23 specialized roles, slash commands, and founder-focused workflows.

Self-improving cross-platform agent with messaging gateway, memory, subagents, automations, and multi-runtime execution.

Crypto-focused MCP and skills launchpad covering chains, DeFi protocols, and exchange integrations for onchain AI agents.

Open-source AI-powered chess learning and analysis platform with interactive gameplay and commentary features.

Kaspa-focused onchain data and AI agent project with 3D visualization components.

Open-source toolkit for testing, validating, debugging, and scaffolding Model Context Protocol servers, positioned as Postman for MCP.

High-performance JAX-based molecular simulation toolkit with differentiable primitives for CPU, GPU, and TPU.

Cross-platform research skill that searches Reddit, X, YouTube, HN, Polymarket, GitHub, and the web to synthesize recent signal.

Agent skill that lets AI agents trade on Lighter with market data, paper trading, account ops, and live execution.

Deterministic truth-synthesis and settlement runtime for Kalshi-style prediction market verification with low-latency processing and on-chain attestation concepts.

Serverless LangChain.js agent example with MCP integration, APIs, web apps, and Azure deployment for real-world tool calling.

Runtime CLI generator that turns MCP, OpenAPI, or GraphQL endpoints into token-efficient command-line tools with zero codegen.

Community-editable skills repository for Claude Code’s Superpowers plugin ecosystem.

Marketing skills library for Claude Code and AI agents covering CRO, copywriting, SEO, analytics, and growth workflows.

Cross-platform MCP server implementing OpenAI-style computer use actions for mouse, keyboard, scrolling, dragging, and screenshots.

Pairing agent for ML engineers and researchers with autonomous baselining, Kaggle workflows, paper integrations, debugging, and reports.

Multi-agent system that transforms research paper codebases into interactive AI agents and MCP servers with minimal human input.

AI coding assistant plugin for Qt Creator with chat, code completion, inline refactoring, MCP server/client capabilities, and project-aware tools.

Qwen coding-agent CLI and terminal environment whose supported model set includes Qwen3.6-Plus across DashScope and Coding Plan integrations.

Token-optimization CLI tooling around RTK, often discussed alongside Caveman compression modes for Claude Code workflows.

AppLaunchFlow MCP workflow for generating app store screenshots and related capture flows.

Lightweight local AI agent harness written in C with built-in tools, cron jobs, webhooks, persistent memory, and a human-readable soul file.

Claude Code plugin that converts repeatable session workflows into reusable SKILL.md files through an interactive capture flow.

Python framework and registry for packaging modular agent capabilities as installable skills with code, instructions, governance, and schemas.

Design skill and CLI that helps coding agents generate, iterate, and refine UI designs on an infinite canvas using structured design workflows.

Kubernetes-native platform for orchestrating fleets of AI agents with isolated pods, CRDs, workflows, memory, and cluster administration.

Open-source desktop app for managing markdown knowledge bases with git-first, offline-first workflows designed to work well with AI agents.

Open-source structured web research agent foundation from Firecrawl with skills, subagents, browser tools, and deployable templates.

Adaptive multi-agent development workflow system with specialized agents, gated phases, structured artifacts, and QA loop-backs.

Transparent desktop arcade overlay that lets users play retro games while AI coding agents work in the background.

Browser automation and web scraping platform with workflow integrations and MCP-compatible AI tooling.

Local observability layer for Claude Code with multi-session monitoring, context pressure signals, cost trends, and live dashboards.

Email MCP integration for agents, especially Hermes, enabling inbox send/read workflows via MCP.

Natural-language interaction layer for Kubernetes/EKS clusters via MCP-style agent workflows.

Official GitHub MCP server exposing structured tools for repositories, issues, pull requests, and related workflows.

Claude Code plugin that uses Codex CLI for PR review, bug validation, false-positive filtering, and optional fix loops.

Unified MCP server for booking and local services across 75+ services and multiple client capability modes.

Open-source managed agents platform that turns coding agents into teammates with task assignment, progress tracking, autonomous execution, and reusable skills.

Local desktop control panel for OpenClaw with persona switching, skill management, memory browsing, backups, and restore flows.

Personal local AI agent and desktop automation assistant for Mac and Windows, controlled through messaging apps like Telegram, WhatsApp, WeChat, and Discord.

Visual-first reference pack for improving UI detail accuracy in iOS and SwiftUI agent builds across Claude Code, Cursor, and Codex.

Security-focused Claude Code skills pack inspired by Trail of Bits workflows for auditing, vulnerability detection, static analysis, testing, and review.

Local-first wallet and finance stack for AI agents with constrained onchain flows, local key custody, and an integrated MCP layer.

Google Cloud Run based remote MCP deployment offering for managed agent tool servers.

Design skills platform for AI agents with aesthetic and engineering skills, plus installable design-oriented agent workflows.

Google Cloud Run based managed remote MCP deployment path for agent tool servers.

Backend development platform for AI coding agents that exposes databases, auth, storage, functions, and deployment through an agent-friendly semantic layer.

Autonomous operator or action engine product positioned around taking missions and executing them as an agentic system.

CLI and library that turns any MCP server into typed TypeScript clients to dramatically reduce tool-schema prompt bloat.

Official SDK for building durable, long-running agents with tool use, handoffs, and agent workflows.

Universal skills loader for AI coding agents that brings Claude Code style SKILL.md workflows to tools like Cursor, Windsurf, Aider, and Codex.

Autonomous machine-to-machine workflow tooling centered on privacy-preserving zero-knowledge proof switching for sensitive data.

Official Taskade MCP server and OpenAPI-to-MCP codegen toolkit with 50+ tools for workspaces, projects, tasks, agents, automations, and media.

Security-focused Claude Code skills pack for audit, vulnerability detection, testing, and review workflows.

Curated list of Claude skills, resources, and tooling for customizing Claude and Claude Code workflows.

Large Claude Code handbook covering prompt patterns, MCP, skills, plugins, subagents, and workflow guidance.

Official Anthropic-managed directory of high-quality Claude Code plugins.

Agent-native personalized learning assistant project from HKUDS focused on adaptive tutoring workflows.

Official open-source Gemini CLI that brings Google’s coding agent and Gemini capabilities directly into the terminal.

Anthropic’s public repository of Agent Skills demonstrating creative, technical, enterprise, and document workflows for Claude.

Claude Code plugin for capturing session activity, compressing memory, and reinjecting relevant context into future sessions.

Official Python SDK for Claude Code for programmatic integrations and agent workflows.

Prompt engineering guide for Claude workflows with MCP, skills, and advanced prompting patterns.

Embeddable presentation/flipbook viewer with an integrated MCP server for programmatic creation and management by AI agents.

GitHub’s official MCP server for repositories, issues, pull requests, and broader GitHub workflows.

Marketing skills library for Claude Code and AI agents covering CRO, copywriting, SEO, analytics, and growth workflows.

TypeScript example agent that connects Ollama models to multiple MCP servers through an interactive CLI.

Agentic skills framework and software development methodology powering the broader Superpowers ecosystem.

Experimental skills and ecosystem extensions for Claude Code Superpowers, covering new techniques and MCP-oriented tooling.

Secure WordPress-to-LLM connector plugin for AI-powered site management and broader WordPress agent workflows.

Guided software development framework extension focused on creating reusable agent skills and structured learning loops.

TypeScript MCP template for orchestrating multiple AI agents with tools for email, scheduling, and knowledge workflows.

Production-ready Microsoft agent framework for multi-agent orchestration, multi-provider models, and interoperability via standards like MCP and A2A.

Curated Claude Code plugin marketplace in the Superpowers ecosystem.

MCP server and integration path for using Snowflake Cortex Agents from external agent workflows and tools.

WeChat AI agent bridge SDK that connects arbitrary agent backends to WeChat through a simple agent interface and ACP adapter.