Custom agents

Facebook Ad Library MCP server and CLI for Claude Code, Codex and AI agents. 9 tools that read every ad running on Facebook, Instagram, Messenger, Threads and Audience Network, for any advertiser in any country, free and with no API key.

firstmate for Windows: talk to one Claude Code agent, ship with a crew of parallel coding agents. Runs on Git Bash and Herdr.

Turn an agent orchestration protocol into a tool-call gate: no child agents spawn until a contract is locked and a driver loop exists. A Pi extension.

🚀 MCP Hub - World's Most Complete MCP Server Collection

Agent Skill：文档/表格先转 Markdown 再读（MinerU 4.x；Windows + Unix 脚本）

A browser extension for LLM-powered translation using your own API key

AI-agent-controllable 3D director desk: place models, mount actions, compose shots, light scenes, export stills/MP4. One serializable command layer drives UI, host apps and LLM agents (local HTTP+WS bridge). React 18 + three.js, embeddable npm package.

An all-in-one repo of the Fast/AllSet SDK packages, CLI, and Skill

Screenshot anything, instantly — zero-dependency, cross-platform screen capture for humans and agents. CLI, Python library, and MCP server. No Playwright, no browser.

A text language for diagrams an LLM can write, drawn as native, editable PowerPoint shapes a person can finish.

MCP server for Figma design: agents read and write the live Figma file and see a screenshot of what they drew. Works on the free plan, zero dependencies.

Search every past coding-agent session on your machine, from any checkout, in milliseconds. A CLI and an MCP server for Claude Code and Codex CLI. No index, no daemon, no setup.

让 AI 正确地排八字：能算的交给 tianzhi-core，模型只负责讲人话

离线私有化知识库：内网和信创环境（麒麟 + OpenCode）下，用 DeepSeek、Qwen 等本地大模型维护文档知识库。Word、Excel、PDF 自动入库，每个回答都附核对过的原文出处，可溯源、不编造；只需 Python 标准库，不联网。

Intelligent routing layer for AI agents: recommends the best MCP server and LLM for any task, scored on 132+ real benchmark executions.

Artifact-gated kanban workflow for AI coding agents. Gates read files on disk, not the agent's summary: every entry into testing mints a fresh execution id, so yesterday's green report stops counting.

🐾 Codecat: A lightning-fast CLI tool that bundles your entire codebase into a single, clean Markdown file. Perfect for feeding context to LLMs, streamlining code reviews, and creating portable project archives.

Local vibe writer that writes with you.

LLM personality evaluation and monitoring platform

DSH 1.5 上下文治理：折叠工具结果、保持提示词缓存命中、接管宿主可写请求（不再需要 dsh-llm 文件补丁）。仅支持 DSH 1.5，后续会跟进更新。 / Context governance for DSH 1.5: folds tool results, keeps the prompt cache warm, takes over the writable-request path. DSH 1.5 only; newer releases will follow.

52 Claude Code subagents, 44 skills, and the .claude/ project-brain pattern I use in every repo.

MCP server exposing Wallapop marketplace search & categories to any LLM client (Claude Desktop/Code). Published on npm. TypeScript.

Ollux — a local AI assistant runtime for Ollama-powered LLM workflows.

Edward Tufte's visual design principles as agent skills — 38 skills plus a router that picks them for you. Works in Claude Code, Codex, Copilot, Gemini, or any agent that can read a file. Every claim verified page-by-page against all five books.

Find where coding models fail across providers and coding tools.

Homebrew tap for mcp-upgrade

AI 工作交接套装 | 一键生成交接文档，让下一个 AI 无缝接手

360 production-ready agent skills for Claude Code: frontend, backend, AI agents, RAG, video/UGC, marketing, DevOps, security, testing + the skill-router system that runs them. One-command install.

Model Context Protocol server for Unreal Engine 5.7. 101 editor tools — actors, selection, paged enumeration, generalized object reflection (any UObject), viewport + screenshots, PIE control, Content Browser navigation, asset migration, materials, Blueprints, UMG, INI editing. Drive UE from Claude / Cursor / Codex / Windsurf via natural language.

command shell mcp server with additional tools (e.g. for coding)

Portability + coordination layer for multiple AI CLIs: write a skill once and run it in Claude, Gemini, Qwen, iFlow, Qoder, CodeBuddy, Codex, KiloCode, and opencode — and share session context across CLIs.

**YYC³ AI Agent Archive**是企业级 AI Agent 资产平台，以 Monorepo 架构聚合了 14 个核心 TypeScript 包、8 位 AI Family 智能体、831 个技能资产、200+ 插件 以及完整的多智能体协同体系。项目以"言启千行代码，语枢万物智能"为核心理念，构建了从智能体定义、技能注册发现、MCP 运行时、协同编排、沙箱执行、可观测性到国际化支持的完整技术栈。

Virginia ABC product search and store inventory from your terminal — agent-friendly, read-only.

Identify which LLM produced given answers — based on single-token behavioral fingerprints (Bruckner 2026, arXiv:2607.10252). 170 models, 10 tasks, 4 languages.

Prompt a parametric 3D model - with your own tokens. Describe a part. The model writes OpenSCAD, your browser compiles it, and you get an exportable mesh.

MCP server that computes exact game-theoretic answers for LLMs — Nash equilibria, dominance, subgame perfection, and repeated-game thresholds. Written in Rust: no network, no keys, no state.

AdSense for AI conversations. Open-source contextual ad engine for chatbots and AI apps using vector search (Qdrant) and LLM embeddings.

Coding-agent trajectory portability: Claude Code to ATIF v1.7 with fidelity accounting, redaction, and historical observability projection.

Keyless connector templates that let an AI agent call any REST API from its documentation. No server, no SDK, your key stays local. MCP-lite for the long tail of REST.

Nix build system for N64 / ModRetro M64 software, with a Faust DSP bridge

Make AI-generated code easier to trust, review, and maintain with practical rules for clearer changes, better explanations, tested tests, and cleaner contributions.

Ultra-fast, fail-open advisory decisions and verbatim extractive reading view for AI coding agents and CLI pipelines

MCP server for solar asset management: reads SolarNetwork telemetry, finds dated equipment faults, and writes field-ready service reports as PDFs.

Block AI-generated garbage sources from your research. Copy-paste fixes for Claude, ChatGPT, Perplexity. Community-maintained blocklist.

Run local coding agents with full tool access and browser-based monitoring using an uncensored Ollama model on your hardware.

Claude Code plugin + zero-dependency CLI for TypeSafe's Jev: judge items in bulk with calibrated yes/no, pick-one and rubric answers; handle the confident ends in code, review only the uncertain band.

Menu-bar app: PDFs, Office & images → clean Markdown on your Mac. Offline, local, no API keys.

Remote execution node that speaks MCP. One static binary; dials out from behind NAT, no inbound ports, no SSH.

Forge an AI that knows you, remembers, and ascends. Open source, free, yours to imprint.

Harness vertical de automatizacion BIM/Revit: conecta un LLM con Revit via WebSocket/MCP, con foco en eficiencia de tokens

Practical, code-first LLM tutorials covering pretrained inference, embeddings, RAG, LoRA and full fine-tuning, dataset evaluation, and training a small language model from scratch with PyTorch and Hugging Face.

Facebook & Instagram public data for AI agents: pages, posts, groups, Ads Library, Marketplace, profiles, reels. 12 read-only tools. npx or hosted at mcp.socialapis.io/mcp

Praukron: shared project memory and governance for people and AI agents. Phases, tasks, decisions and acceptance contracts as versioned Markdown, compiled into deterministic project state.

An MCP server that lets AI agents create and manage LighterPack packing lists — add gear, track weights, and organize categories through natural conversation.

MCP server for Ableton Live 12+: read and edit the whole Set (clips, notes, devices, automation, routing, browser) through addresses, plus composition tools, batches and undo. Node/TypeScript + a Python Remote Script.

Deterministic multi-tradition astrology engine — Western tropical, Vedic/Jyotisha sidereal, Chinese BaZi. Zero-dependency pure-Python ephemeris, 19 modes, arcsecond-accurate (validated vs NASA JPL). CLI + API + MCP.

Turn any file into a CLI from TypeScript/JavaScript exports

MCP server exposing the X (Twitter) API v2 to LLM agents: 42 tools for posts, users, lists and media, over OAuth 1.0a. Written in Go.

Make every Android app agent-ready in 60 seconds. No code changes. No SDK.

Clean-room Mojo kernels as drop-in accelerators for popular Python/TS libraries — measured 70x-8,700x with bit-exact parity and pure-language fallbacks. bm25, Fuse.js fuzzy search, cclib electron-density grids. Siblings: thyn-ai/algenta-sdk, thyn-ai/algenta-integrations.

the top benchmarked pstack port for Claude Code: contains playbooks, skills and principles, enforced by a hook at every action

A TRPG Game Master that never forgets or contradicts. Outcasts Lorekeel uses a cloud LLM for rich storytelling, backed by a deterministic Rust engine as the source of truth for all game states. By enforcing strict state mutation rules, it runs consistently even on cheap or local models. Built with Tauri and Vue.

Local LLM does Claude's bulk work: reads logs and files, researches the web, writes test-gated code.

🤖 Monitor $BTB on Twitter, engage with the community, and provide real-time information using AI-powered responses for BTB Finance.

Zaru, an AI assistant in your terminal: it reads and edits files, runs commands and fetches pages, asking first.

Portable WPE WebKit runtime for AI agents with native input, screenshots, ordered JSONL, and reproducible benchmarks.

STRATA-Bench: Spatial-Temporal Retrieval, Alignment, Transparency & Audit — a benchmark for AI agents on fragmented geographic and longitudinal market intelligence.

The merge queue built for AI agent swarms — batches, bisects, and proves every landing.

EliaAI – AI Agent System for macOS

POK-Agent: a Windows computer-use AI agent. A large model plans (System 2) and a small fast model acts on screen (System 1: JEV or local Laya). Repeated tasks become muscle memory, so the big model is called less every time. Local-first, voice and vision, Rust + Tauri, open source.

An open, MIT-licensed AI coding agent in Python. Read the source, fork it, self-host it. A Mvge (agent) casts Spells (tools) against models served by Realms (providers), which are pluggable; the OpenRouter Realm ships as a Rune. Conversations persist as Tomes (sessions) in JSONL. Terminal and desktop interfaces.

Assembled context for DeepSeek Harness: a context tree over the session surface with per-node assemble modes (full / key / off), agent-authored presets, and a panel that decides what the model actually sees.

从 B站海量视频里为一个议题取数：双轴信源评级（A–F × 1–4）、冲突单列、三类缺口声明、检索协议可复现。MCP server + 方法论 skill。

把团队文档变成 Agent 的 SOP，只用你批准过的只读工具查真实数据 · Turn team docs into agent SOPs, executed with admin-approved read-only tools

AI event probability prediction engine

专为 AI Agent 设计的本地游戏汉化与文本工程工具链（内置 Claude Code / Codex / DeepSeek Harness Skill）。原生支持 Unity (IL2CPP/Mono/UnityPy)、RPG Maker、Godot、Ren'Py、WOLF、TyranoScript | AI Agent Game Localization & Translation Toolchain.

AstrBot 的插件与补丁合集：自建网页检索（走本机 Edge，不依赖搜索 API）、把 QQ 收藏表情发进会话、脱离 ComfyUI 的 SDXL 生图后端（多底模/多画风/两档采样器/画幅上限全部可配）、看头像与历史回溯

Batch Code Reasoner — Non-interactive LLM job runner

Nemotron agent for diagnosing RL training failures and verifying recovery in Isaac Lab.

Skyward — the first open world AI agents can log into and build, forever. WebGPU/Three.js, open MCP agent ingress. Source-available (FSL-1.1).

A collection of AI-friendly architecture guides based on Modular Monolith, Feature-Based Design, and pragmatic Clean Architecture, promoting low coupling, simplicity, and maintainability across different software projects.

A live, contamination-resistant benchmark for evaluating LLM agents on U.S. macroeconomic nowcasting.

CSVX is an open, plain-text spreadsheet format for the AI era: human- and AI-readable CSV-shaped tables with types, formulas, formatting, comments, structured annotations, multi-line cells, and sheets.

Free Claude MCP server for TabICL — tabular ML predictions, explanations and shareable reports

Fine-tuning de Mistral 7B Instruct (QLoRA, Unsloth) pour écrire à la manière de Victor Hugo, Baudelaire et Verlaine. Notebook Colab, jeu de données Project Gutenberg, évaluation par perplexité.

Turn data files (CSV/Excel/JSON/JSONL/TSV) into a validated database, then operate it in natural language — pluggable methodology, quality gates, migration ledger, visual delivery

R² Flow: Recursive Self-Improvement via Recursive Skill Evolution — training code

AI agents that rebuild a web application from its observable behavior — clean-room, with a human-approved barrier and a hash-chained audit trail.

Talk2Code — turn one sentence into a runnable web app. Multi-agent PM/Coder/QA pipeline with real-browser (Playwright) acceptance tests and a delivery gate.

Stop burning 30% of your agent's context on unused skills. 819 curated superpowers across 16 IDEs. One autonomous hive equipping the exact skills you need in 9ms. Take total command—and watch yourself build miracles with your agents as they turn your raw ideas into effortlessly shipped software.

Multi-Agent Natural Language to SQL System: a five-agent NL→SQL pipeline over DuckDB with AST validation, grain-aware fan-out checks, row-level security, PII masking, a hash-chained audit log and a 137-case eval suite with an ablation ladder gated in CI.

Agentic framework that ports software to RISC-V. Native riscv64 builds, automatic repair, verified binaries and a replayable recipe for each package.

本地优先的多智能体非完全信息博弈基座：主持人控制台、席位令牌视野、事件驱动 CLI、持久快照与可信游戏 Mod

Four composable Agent Skills that let a coding agent drive, inspect, debug, and profile iOS apps — with evidence instead of guesses. Works in Claude Code, Codex, Cline, and any Agent Skills runtime.

need many characters different thinking different knowledge base, scrapping, different tools for automation, as I already have freellm-business-agentic-os, but that lacks many of my purposes, so lets built a proper one, to use for automation puposes

🚀 SiliconFlow-AI: Autonomous RTL-to-GDSII Physical Design Framework uniting LLM Agents with native EDA engines (Verilator, Yosys, OpenSTA, OpenLane) on SkyWater 130nm PDK.

Automated KYC & AML Screening Agent — ReAct agent for PEP/sanctions screening, UBO extraction, adverse media monitoring

An always-on founder intelligence system.

Agent skill: turn GitHub issues into PRs the superpowers way — bootstraps superpowers if missing, runs the full interactive flow, one branch + PR per issue

An agentic OS in your browser: the chat is the interface and apps get generated on demand as HTML. Local model via WebGPU, zero install.