Explore

GoRaven is a self-hosted, team-oriented AI agent platform in Go that gives each user an isolated agent workspace with tool calling, code/shell execution, MCP integration, RAG knowledge, and an installable skill marketplace. One Docker command runs it with OpenAI, Claude, DeepSeek, Qwen, or Ollama backends.

AI Agents | Go · ai-agent · multi-agent

Git-native persistent project memory for AI coding agents, storing decisions and facts as plain-text OKF v0.2 markdown bundles. Serves them over an embedded MCP server with sub-300µs in-memory BM25 search and progressive disclosure, cutting context token bloat ~80% with no databases or API costs.

AI Agents | Go · agent-memory · persistent-memory

Rust MCP server and CLI that gives agents web fetch, keyless search, crawl, and PDF/OCR extraction from one binary. It uses real Chrome BoringSSL TLS plus stealth and learning layers to get past bot walls, with no API keys required.

MCP | Rust · mcp-server · web-fetch

Windows-first runtime that lets you speak to NPCs in unmodified games: speech-to-text, LLM dialogue with vector-store memory, TTS plus SadTalker lip-sync injected back into the game, and webcam emotion awareness.

AI Agents | Rust · llm · npc

Unofficial zero-dependency CLI that reads and writes CapCut/JianYing draft JSON directly to edit projects: subtitles, timing, speed, templates, and long-form-to-shorts cuts. Ships an agent skill and Claude Code plugin so LLM agents can drive edits, with no API or server.

AI Tools | JavaScript · capcut · jianying

StoryMaker is a diffusion personalization method and inference/training pipeline that keeps characters consistent across text-to-image generations, preserving faces, clothing, hairstyles and bodies. Built on SDXL with IP-Adapter, LoRA and InsightFace encoders.

AI Tools | Python · text-to-image · character-consistency

Python agent that turns natural-language prompts into Wyckoff volume-price analysis, A-share/HK/US funnel screening, AI research reports, and portfolio risk control. Ships CLI, web, desktop, and an 18-tool MCP server with layered memory, backtesting, and scheduled daemon runs.

AI Agents | Python · wyckoff · trading-agent

NestJS module that turns decorated controllers into Model Context Protocol servers, exposing tools, resources, and prompts over Streamable HTTP or STDIO. It reuses NestJS DI, guards, and pipes, and supports OAuth auth, elicitation, and dynamic capability registration.

MCP | TypeScript · mcp-server · nestjs

Official spec repository for the MCP Skills extension (io.modelcontextprotocol/skills, SEP-2640), plus the working group's decision log, rationale, and implementation list. Read it to implement or evaluate serving and discovering agent skills over MCP; it is documentation, not a runnable server.

MCP | MDX · skills · specification

Horizon is a Rust desktop app that places terminals, coding-agent TUIs (Claude Code, Codex, Gemini, Grok), live browsers and git panels on an infinite canvas grouped into workspaces. It ships an MCP browser-automation server agents can share, plus session persistence and token-usage tracking.

AI Coding | Rust · terminal-emulator · infinite-canvas

Unreal Engine 5.8+ plugin that expands the engine's native MCP server with editor toolsets and ~36 agent skill packs, letting Claude Code, Cursor or Copilot author Blueprints, landscape, Niagara, animation, UMG and AI behavior, plus profile CPU/GPU frames in-editor.

MCP | C++ · mcp-server · unreal-engine

Zero-runtime-dependency TypeScript SDK for production AI agents: durable resumable runs, long-term memory, hybrid RAG, MCP tool calling, approvals and swarm orchestration over one streaming API for Claude, GPT, Gemini, Grok, Mistral and DeepSeek on Node, Bun, Deno and edge.

AI Frameworks | TypeScript · agent-framework · durable-execution

yutu is a Go CLI, MCP server, skill pack and Gemini-powered agent for YouTube: it automates uploading videos, optimizing metadata/SEO, and managing playlists, comments and channel settings. Works standalone, in CI, or attached to Claude Code, Codex, Cursor and VS Code.

MCP | Go · youtube · mcp-server

Buoy adds a floating suite of debugging tools to React Native and Expo apps, with a desktop dashboard and an MCP server that lets AI agents inspect app data and perform supported actions.

MCP | MCP server · React Native

Curated awesome list for Google's Agent2Agent (A2A) protocol: official specs, SDKs, framework integrations, sample agents/servers, tools and tutorials. Best as a starting index for developers building interoperable multi-agent systems.

AI Agents | a2a · agent2agent

Zero-dependency browser non-linear video editor whose whole timeline is one JSON document. Ships an MCP server (and REST API) so Claude Code, Claude Desktop or any agent can cut, grade and export video while the UI live-reloads.

MCP | JavaScript · video-editing · ai-agent

MCP server that gives LLM clients read/write access to an Obsidian vault over the Local REST API plugin: 14 tools and 3 resources cover note CRUD, section/block patching, frontmatter, tags, and text/JSONLogic/Omnisearch queries. Runs via STDIO or Streamable HTTP with folder-scoped permissions.

MCP | TypeScript · mcp-server · obsidian

Arcjet's JS/TS SDK adds runtime security to AI apps and agents: prompt-injection detection, tool-call authorization, PII redaction, bot protection, and rate limiting. Framework adapters (Next.js, Node, Bun, Deno, etc.) plus @arcjet/guard for MCP servers and background jobs.

AI Tools | TypeScript · security · prompt-injection-detection

A cross-platform AI agent app for Android, iOS, and macOS, built on the Pi framework. It supports tool calling, Pi extensions, MCP integrations, and an Alpine Linux environment for running shell commands and tools.

AI Agents | Kotlin · AI agent · local AI

MCP server plus Figma plugin that streams live Figma document data over WebSocket, bypassing Figma API rate limits. Exposes ~40 read and write tools so agents like Cursor or Claude Desktop can inspect designs and make gated edits.

MCP | TypeScript · mcp-server · figma

A Laravel package that exposes Eloquent repositories as JSON:API endpoints and MCP tools from a shared definition. Laravel policies, validation, search, and filtering apply across both interfaces.

MCP | PHP · MCP server · Laravel

MCP server exposing 45 tools that let Claude discover, analyze, and trade Polymarket prediction markets on Polygon, with WebSocket monitoring, portfolio/P&L tracking, and configurable risk limits. Python, requires wallet credentials for live trading.

MCP | Python · mcp-server · polymarket

A FastMCP server that gives MCP clients tools for stock data, technical analysis, screening, portfolio tracking, and optional strategy backtesting and LLM-assisted market research. Supports STDIO and Streamable HTTP.

MCP | Python · stock analysis · financial data

Adapter for hosting MCP servers on Vercel and Next.js routes.

MCP | TypeScript · Vercel

MetaHarness scaffolds customizable, repo-aware AI agent harnesses from a GitHub URL or blank slate. Generated harnesses can include a CLI, MCP server, project-scoped memory, skills, governance, and release verification for supported agent hosts.

AI Agents | TypeScript · agent-scaffolding · agent-harness

Framework-agnostic HTTP adapter that turns an MCP server definition into a Web-standard fetch handler mountable in Next.js, Nuxt, SvelteKit, and Hono. Supports the 2026-07-28 MCP spec with legacy Streamable HTTP fallback, plus OAuth resource-server helpers.

MCP | TypeScript · mcp-server · http-adapter

Kody is a multi-user personal assistant platform exposing OAuth-protected MCP endpoints. It gives each user isolated memories, secrets, code packages, and automations, with a compact search and Code Mode execute interface.

MCP | TypeScript · MCP server · AI assistant

A collection of installable skills that guides AI coding agents through Scenario MCP workflows for image, video, audio, and 3D generation. Includes expert skill teams for operating Blender, Maya, ZBrush, Unreal Engine, and Unity.

Skills | Python · agent-skills · generative-ai

Local-first runtime security for AI agents: hooks and MCP proxies that classify shell commands, file access, package installs and tool calls, then allow, block or request approval. Also ships plugin-scanner, a CLI/CI scanner for agent plugins, skills and MCP servers.

AI Tools | Python · agent-security · mcp-security

Self-hostable infrastructure for connecting AI agents with people in Slack, Teams, Discord, Telegram, and Mattermost. It provides shared rooms, agent protocols, and workflow coordination without supplying agents or models.

AI Agents | Python · multi-agent · agent orchestration

A local AI video production workbench that turns Chinese scripts and reference audio into narrated whiteboard animations or dynamic infographics. It connects to text and image model APIs plus an IndexTTS service, then renders the finished video as MP4.

AI Tools | Python · AI video generation · whiteboard animation

An MCP server for jailbroken iPhones that exposes 46 tools for device control, app management, accessibility inspection, OCR, file access, and logs. Connect compatible AI clients to automate and debug a physical iOS device.

MCP | Objective-C · iOS · jailbroken iPhone

An MCP server that connects AI agents and coding assistants to SonarQube Server or Cloud for code quality and security information, including analysis of code snippets.

MCP | Java · MCP server · SonarQube

Dexto is a TypeScript agent harness with a CLI, web UI, and SDK for building stateful agents. It includes a coding agent, multi-agent delegation, persistent sessions, tool approvals, and MCP client/server support.

AI Agents | TypeScript · agent harness · coding agent

.NET SDK for building AI agents and workflows with 30+ provider connectors, MCP and A2A support, vector DB integrations, multimodal IO, and a graph-based agent orchestration API. Works with local runtimes like vLLM, Ollama and LocalAI.

AI Frameworks | C# · dotnet · csharp

BioMCP is a Rust CLI and MCP server exposing one command grammar over ~30 biomedical sources: PubMed, ClinVar, ClinicalTrials.gov, OncoKB, Reactome and more. Plug it into Claude Code, Codex or any MCP client for evidence-oriented gene, variant, drug, disease and trial queries.

MCP | Rust · mcp-server · model-context-protocol

CLI and local dashboard that runs staged LLM code reviews over a Git diff or a full codebase using TypeSafe Jev, scoring correctness, security, reliability and test gaps. Orchestration and thresholds live in code; findings are review prompts, not proof of defects.

AI Coding | TypeScript · code-review · ai-code-review

Rust MCP server giving agents a local, encrypted, deterministic memory layer: writes are gated and merged, unused memories fade via FSRS-6, and backfill retrieval traces a failure backward to the earlier decision that caused it, with receipts.

MCP | Rust · agent-memory · mcp-server

SigMap builds a deterministic, offline map of repository files and symbols for AI coding assistants. Its CLI and MCP server support code search, compact context generation, and verification of AI answers against real code.

AI Coding | JavaScript · code-context · code-grounding

A Go gateway for AI coding agents that routes requests to 44 model vendors through one endpoint. It monitors agent activity, grades API authenticity, enforces per-agent permissions, and records prompts and usage.

AI Coding | Go · AI coding agents · LLM gateway

A desktop app that turns selected clipboard captures and saved links into an AI-organized personal wiki. It supports knowledge-base Q&A, graph visualization, insight reports, and MCP integration, with data stored locally in SQLite.

AI Tools | Rust · AI knowledge management · personal wiki

An MCP server that lets compatible AI clients deploy applications to Google Cloud Run and manage services, projects, and logs. Supports local or remote deployment and authentication with Google Cloud credentials or OAuth.

MCP | JavaScript · MCP server · Google Cloud

FLUJO is a local-first, MIT-licensed Next.js app for visually building multi-agent workflows as graphs, wiring in MCP servers and multiple LLM providers. It ships a debugger, automation triggers, and OpenAI-compatible plus MCP proxy endpoints so agents can be called from other tools.

AI Agents | TypeScript · multi-agent · agent-builder

Native Swift/SwiftUI macOS autonomous agent that wires 21 LLM providers plus on-device Apple Intelligence into a self-verifying task loop. Drives any Mac app via the Accessibility API, runs privileged shell, edits and builds Xcode projects, supports MCP servers, voice control and iMessage remote control.

AI Agents | Swift · macos · swiftui

Official Redis MCP Server exposing Redis as a natural-language toolset for MCP clients: strings, hashes, lists, sets, sorted sets, streams, pub/sub, JSON, plus vector index/search. Ships as a PyPI package, Docker image and uvx entry with EntraID auth.

MCP | Python · mcp-server · redis

Surf is a zero-config CLI that lets AI agents control Chrome and other browsers via Unix-socket commands, no MCP relay or API keys required. It offers page reading, screenshots, network capture, remote Tailscale access, and also ships an MCP server.

AI Tools | TypeScript · browser-automation · chrome

A curated directory of APIs for building autonomous AI systems, organized into agents, AI models, and MCP servers. Browse provider links and descriptions to identify services for agent workflows, model-powered features, and tool integrations.

AI Agents | JavaScript · API directory · MCP servers

Self-contained 3D Gaussian Splatting trainer: photos/video to splat to textured mesh in one binary, with Vulkan (NVIDIA/AMD/Intel/Apple) and CUDA backends. Built-in SfM, SAM masking, 360/fisheye support, and quantized training for 10M Gaussians in 8GB VRAM.

AI Tools | C++ · 3d-gaussian-splatting · 3d-reconstruction

Marmot is a self-hosted Go data catalog that indexes tables, topics, queues and APIs, then serves that metadata to AI agents through an MCP server. Useful if you want LLM tooling grounded in real, governed data context.

MCP | Go · data-catalog · metadata

EMILIA provides an open protocol and enforcement boundary for consequential agent actions. Gate checks exact-action authority, admits covered provider attempts, and creates verifiable evidence; it is effective only on fully mediated paths with authoritative shared state.

AI Agents | TypeScript · agent security · authorization

A local-first MCP server that lets AI clients search, read, and manage a Zotero library. It uses a bundled C++ BM25 search core for PDF and passage retrieval, with TypeScript and Python shells and stdio or Streamable HTTP transports.

MCP | TypeScript · MCP server · Zotero

agentty is an MIT-licensed terminal coding agent written in C++26, shipped as one ~17MB static binary with millisecond startup and no Node/Python runtime. It supports many providers (Anthropic, OpenAI, Groq, Gemini, Ollama, OpenRouter, etc.), sandboxes shell commands, and ships a fully local hybrid BM25 + embedding Gra

AI Coding | C++ · terminal · cli

Napkin is a Claude Code (and Codex) skill that keeps a per-repo markdown scratchpad at .claude/napkin.md, where the agent records its mistakes, your corrections, and preferences each session and reads them back at start. It gives coding agents lightweight persistent memory so recurring errors drop off after a few sessi

Skills | claude-code · agent-memory

Memoir is a Python memory system for AI agents with hierarchical semantic paths, search, and Git-like branching, commits, and rollback. It includes a CLI, MCP server, and integrations for coding and assistant agents.

RAG | Python · agent memory · semantic search

IronCurtain is a TypeScript runtime that keeps autonomous AI agents untrusted: it compiles a plain-English constitution into a deterministic policy, then enforces allow/deny/escalate on every MCP tool call inside a V8 or Docker sandbox. Also runs multi-agent workflows with human gates.

AI Agents | TypeScript · agent-security · sandboxing

Claude Code skill that runs /tech-debt-audit to produce a whole-repo, file:line-cited tech debt report (TECH_DEBT_AUDIT.md) with severities, effort estimates, top priorities, and a repeat-run baseline mode. Installs as a single SKILL.md.

Skills | claude-code · agent-skill

OpenRig is a TypeScript CLI and harness for running Claude Code and Codex together as persistent, coordinated agent teams. Define team topologies in YAML, manage sessions through tmux, and enable agents to communicate, hand off tasks, and review work.

AI Agents | TypeScript · multi-agent · agent-orchestration

MCP server that exposes dbt project discovery, semantic-layer queries, SQL execution, CLI commands, job administration, and code generation as tools for AI clients.

MCP | Python · dbt · data analytics

Rust MCP server and CLI that gives any MCP host full control of a Linux desktop: AT-SPI accessibility trees, screenshots, semantic clicks and typing, and compositor-aware window targeting across GNOME, KDE, Hyprland, i3 and COSMIC, Wayland-first with ydotool and X11 fallbacks.

MCP | Rust · computer-use · linux

Wayland is a local-first desktop AI agent (Electron + Rust engine) that acts as a command center for Claude Code, Codex, Gemini and other AI CLIs. It adds persistent memory, skills/workflow libraries, self-assembling agent teams, MCP/ACP integration, scheduling and sandboxed tool execution on your own keys. AGPL-3.0.

AI Agents | TypeScript · desktop-agent · local-first

Self-hosted LLM gateway and tier-routing proxy for Claude Code, Cursor, Codex, and other AI coding tools. Routes requests across local and cloud providers, with semantic caching, prompt compression, and MCP tool support.

AI Tools | JavaScript · LLM gateway · model routing

An agent skill (Claude Code / Codex) that turns a repo's recent commits into a code-rendered software update promo video: storyboard and copy, real product components, motion, original synthesized music and SFX, then FFmpeg export plus an editable project.

Skills | Python · agent-skill · claude-code

An MCP server that lets AI clients manage Coolify deployments, infrastructure, configuration, and diagnostics. Supports local stdio or remote HTTP connections, multiple Coolify instances, and safeguards for destructive operations and secrets.

MCP | TypeScript · Coolify · infrastructure

An agent skill for creating product launch films and feature demos as continuous camera takes. It includes motion and UI libraries, rendering and audio tools, and automated checks for continuity, pacing, and other video issues.

Skills | Python · agent-skill · Claude

A JavaScript CLI for reducing Claude Code context usage by organizing project documentation, measuring token costs, and pruning or compressing CLAUDE.md. Includes hooks for token monitoring and selectively loading relevant project knowledge.

AI Coding | JavaScript · Claude Code · token optimization

TypeScript knowledge-base engine for the pi agent that ingests URLs, PDFs and markdown into an interlinked, Obsidian-compatible wiki, then serves it over MCP. Adds search, embeddings, linting and layered personal/project vaults so agent memory compounds instead of resetting each session.

RAG | TypeScript · knowledge-base · obsidian

A collection of installable skills that teaches AI coding agents workflows for creating motion graphics, animation, and video. Includes render-and-verify guidance for tools such as Remotion and Manim.

Skills | HTML · AI agent skills · motion graphics

An agent skill that loads a curated color science knowledge base (~286K words across 144 references) into coding agents. Covers color spaces, OKLCH/OKLAB, palette algorithms, WCAG/APCA accessibility, pigment mixing and historical color theory, plus eval prompts for review.

Skills | agent-skill · color-science

A local-first dashboard for monitoring Claude Code and Codex sessions across one or more machines. It tracks tokens, cache activity, models, and estimated costs, with an optional self-hosted team hub and Prometheus metrics.

AI Tools | JavaScript · AI coding agents · observability

Agent skill pack (Claude Code / npx skills format) bundling 450+ platform design rules distilled from Apple HIG, Material Design 3, and WCAG 2.2. Each platform skill (iOS, iPadOS, macOS, watchOS, visionOS, tvOS, Android, Web) lets agents review, audit, or generate compliant UI code.

Skills | agent-skills · claude-skills

MiniSearch is a self-hosted private search engine: SearXNG aggregates web results, a local cross-encoder reranks them, and an AI assistant writes cited answers using either in-browser LLMs (WebGPU/CPU via wllama) or an OpenAI-compatible backend. Ships as one Docker container, no API key or telemetry required.

RAG | TypeScript · web-search · in-browser-inference

An MCP server that records sequential reasoning steps with history, branching, revisions, and optional validation of model-authored tool plans. It is a scratchpad plus validator, not a tool-discovery engine, and redacts prompt-injection-like text before storing.

MCP | TypeScript · model-context-protocol · sequential-thinking

A Python agentic personal knowledge system that stores portable context in Markdown and uses RAG, structured reasoning, and governed agents to support decisions across LLM providers. Includes an MCP server and local-first workflows.

AI Agents | Python · local-first · persistent memory

A cross-platform desktop office suite with AI document editing and context-aware agents for documents, spreadsheets, slides, and PDFs. Supports common Office formats, local PDF conversion, and user-provided model API keys.

AI Tools | TypeScript · AI office suite · document editing

TypeScript reference architecture enforcing an authorization boundary for AI agents: agents propose immutable intents, an independent policy engine returns ALLOW/ESCALATE/BLOCK, verified human approval issues a single-use execution grant, and a SafeExecutor runs only granted actions with verifiable audit dossiers.

AI Agents | TypeScript · agent-permissions · agent-safety

DSH Desktop is an Electron desktop distribution of the DeepSeek Harness agent runtime: it bundles official DSH web UI, sessions, MCP tools and a plugin marketplace with file, Git, terminal, worktree, task-board and Office artifact bundles.

AI Agents | JavaScript · ai-agent · agentic-workflows

A local-first Go CLI and MCP server that builds and maintains a Git-versioned codebase and database-schema index for coding agents. Supports persistent repository context across sessions and integrates with Claude Code, Codex, Cursor, and OpenCode.

MCP | Go · MCP server · coding agents

fauxnix is a deterministic bash→PowerShell translator for Windows, shipped as an MCP server and CLI for AI agents. Agents keep writing bash (ls, grep, sed, curl, kill); it translates 109 commands to native PowerShell with GNU-style output and bash-style errors — no VM, no WSL, no LLM calls at runtime.

MCP | TypeScript · mcp-server · bash

Self-hosted MCP gateway that turns 191 pre-built adapters plus any REST, SOAP/WSDL, GraphQL, SQL or MCP system into MCP tools for Claude, ChatGPT, Copilot and Gemini, with per-tool response shaping, RBAC, SSO/SCIM, audit logs and an optional knowledge graph for cross-system chaining.

MCP | TypeScript · mcp-gateway · mcp-server

Swift-native agent runtime for building AI agents with type-safe @Tool macros, Apple Foundation Models on-device inference, composable sequential/parallel/routed workflows, memory, guardrails, streaming, MCP bridging, and OpenTelemetry tracing. Ships as a Swift Package for iOS, macOS, and Linux.

AI Frameworks | Swift · agents · multi-agent

Claw Orchestrator wraps Claude Code, Codex, Antigravity, Grok, OpenCode or custom CLIs as persistent programmable sessions and adds multi-agent councils, autoloop Planner/Coder/Reviewer runs, MCP/ACP servers, and an OpenAI-compatible endpoint. A TypeScript runtime for orchestrating coding agents from CLI or any editor.

AI Coding | TypeScript · claude-code · codex

Unity CLI Loop is an MCP server plus CLI and skill pack that lets LLM coding tools drive a Unity project: compile, run tests, read logs, edit scenes, simulate input, screenshot the Game View, hot-reload and set pause-points. Aimed at autonomous AI-driven Unity dev loops.

MCP | C# · unity · ai-agent

A Rust library for embedding multi-agent applications, with dynamic agent spawning, tool integration, and journaled workflows scripted in Rhai.

AI Frameworks | Rust · multi-agent · agent-runtime

A local-first workbench and chat interface for running Claude Code, Codex, and OpenCode on your own machine. Connect through a browser or messaging apps to start, steer, and monitor coding-agent sessions remotely.

AI Coding | Python · coding-agents · local-first

A stdlib-Python linter that scores copy out of 5 on five AI-writing tells and exits red below 5/5, plus a cleanse step that routes the draft to a rival model family, then re-lints. Ships as an agent skill (SKILL.md) and a CI build gate.

Skills | Python · ai-writing-detection · slop-detection

A macOS and Windows utility that injects custom themes into the Codex/ChatGPT desktop app, with skin management, local token statistics, and an optional theme-generation skill. It uses local Chromium DevTools Protocol injection rather than modifying the official app bundle.

AI Tools | JavaScript · Codex · ChatGPT desktop

An MCP server that gives coding agents persistent project knowledge stored as a SQLite concept graph. Agents can search, add, and update concepts; a web explorer visualizes the graph, and CLI tools support merging databases.

MCP | TypeScript · MCP server · coding agents

Local-first MCP server that reads an Apple Health CSV export into DuckDB and exposes SQL tools (schema discovery, analytical queries, health reports) to MCP clients like Claude Desktop. No uploads; data stays on disk.

MCP | TypeScript · mcp-server · apple-health

自托管的 AI 多智能体投研终端,覆盖 A股/港股/美股行情、深度分析、龙虎榜、风险扫描与走势预测。用 SKILL(Anthropic Agent Skills 格式)承载分析方法论,支持自带大模型 key 和自接 MCP 数据源,对话与持仓全部存本地。

AI Agents | Python · multi-agent · fintech

An X/Twitter automation toolkit with an MCP server, CLI, scrapers, and browser scripts. AI agents can access social data and perform account actions, with optional approval gates and action limits.

MCP | HTML · MCP server · X

An MCP server and OpenAPI service for controlling Proxmox VE. It exposes VM, LXC, snapshot, backup, storage, and cluster operations to AI clients and automation tools, with access controls and guarded command execution.

MCP | Python · Proxmox VE · virtualization

Bilingual (EN/ZH) 23-chapter AI Agent textbook covering tools, memory, planning, RAG, context engineering, agentic RL, multi-agent systems, evaluation, security, and deployment, plus a small runnable reference agent with MCP server and offline tests.

AI Agents | HTML · tutorial · book

Open-source catalog of 148 web design styles for AI-generated UIs. Each style ships design tokens, component recipes, Tailwind constraints and prompts, delivered to coding agents via an MCP server, Agent Skills, Cursor/Claude rules and shadcn registry exports.

Skills | TypeScript · design-system · design-tokens

TDA is a Java thread dump analyzer with a Swing GUI, VisualVM/JConsole plugins, and an MCP server mode exposing parse_log, check_deadlocks, find_long_running and virtual-thread tools so AI agents can diagnose JVM deadlocks, pinning and performance issues headlessly.

MCP | Java · thread-dump · jvm

MCP server that bridges AI agents to TouchDesigner's WebServer DAT, exposing 14 tools and 3 prompts to create, edit, query nodes, run Python scripts, and capture TOP images. Ships a .tox component, Docker setup, and version-compatibility checks; TypeScript, MIT.

MCP | TypeScript · mcp-server · touchdesigner

Claude Code and Codex plugin that runs the Ralph loop on top of a structured spec workflow: research, requirements, design, and tasks are written to disk, then executed task-by-task with fresh context and resumable state.

AI Coding | Shell · claude-code · claude-code-plugin

Self-hosted AI automation platform that turns chat instructions into readable workflows, scoped agents, and small team apps. Includes agent auditing, built-in data tables, and optional browser and web-search services.

AI Agents | Java · AI automation · workflow automation

Automates EasyEDA Pro through typed schematic and PCB actions exposed as a CLI, agent skill, and stdio MCP server. Agents can create and validate circuit designs, manage component libraries, and inspect design-rule checks.

AI Agents | Go · EasyEDA Pro · EDA automation

Recipe and scripts for online iterative RLHF: SFT, reward modeling, vLLM response generation, reward annotation, and iterative DPO training loops. Reproduces LLaMA3-8B alignment comparable to Llama3-8B-Instruct using only open-source data.

AI Frameworks | Python · rlhf · dpo

SMG is a Rust LLM gateway that unifies OpenAI/Anthropic/Gemini and self-hosted engines (vLLM, SGLang, TensorRT-LLM, TokenSpeed, MLX) behind one API. It adds KV-cache-aware routing, gRPC pipelines, MCP tooling, multi-tenancy and observability for large-scale inference deployments.

AI Frameworks | Rust · llm-gateway · inference-routing