Custom agents

Generate conventional commit messages from your staged git changes using Model Context Protocol (MCP)

A tiered team of engineering agents and portable workflow skills for AI coding tools.

Decentralized AI

Trade · Learn · Evolve —— A 股模拟交易员，每一次交易都成为下一次判断的依据 | An A-share paper trader that turns every trade into evidence for its next decision

中标狗：AI Agent 驱动的标书生成桌面应用，从读标、评分废标检查、素材复用到 Word 出件。

在手机上独立运行完整 DeepSeek Harness（DSH）的开源实现。功能已验证，UI 仍在优化中。

A curated collection of powerful agentic skills for automation, development, reverse engineering, productivity, and AI-assisted workflows — designed to extend Claude into a smarter autonomous engineering companion ⚡

Run AI agents on one server. Execute tools across every machine you own.

RAG pipeline with embeddings + semantic retrieval

MCP server for Yellow Pages: search local businesses by keyword and location, read one listing in full with services, brands and payment methods. 1,000 free credits every month.

pochete-toolkit é o conjunto de ferramentas feito pra quem codifica de forma agêntica com claude code em ambiente enterprise.

A highly customizable AI companion for Telegram. Create digital clones with unique personalities, voice, and vision directly from Google Colab.

LLM-led e-commerce content review with rule-based safeguards, tenant-isolated MySQL, Redis workers, and auditable publication decisions.

MCP server for Bing search: organic results, the ad block and the Copilot answer, with geo, market and language targeting. No Azure subscription. 1,000 free credits every month.

Modular node-graph DAG executor for latent image & video workflows (ComfyUI style)

Edit a running web app from inside the page: pick elements, screenshot, preview changes, approve one, and it lands in your source.

Multimodal OCR table grid & merged-cell reconstruction into Markdown (MinerU / Marker)

NyatBot — a Telegram group-chat agent on the path from chatLLM to AGI

Kubernetes troubleshooting workflow

A read-only weekly audit for Codex automations and personal skills, with approval-gated fixes.

Lumpcode : The Open Source CLI and Library for loop engineering

A platform-agnostic, file-based multi-agent engine. Install 20 specialized AI personas directly into your repository. They plan, design, build, review, test, and ship — with structured memory that persists across sessions and two architectural moats no other framework combines.

AI assistant skill files for the HyperTracker API — cohort analytics, order flow, liquidation risk, and leaderboard data from Hyperliquid

A Visual Design that ships with your code. Agents and humans work together on the same canvas, where every change is a diffable JSX stored in a .uidx file inside your repo. Built on OpenPencil and Skia, with MCP built in.

Real-time quadruped robot gait telemetry analyzer & stability evaluator (Unitree / Boston Dynamics)

Give agents capabilities. Keep credentials local. A local relay that gives AI agents, scripts, and automation controlled access to HTTP, SSH, and LLM APIs — without exposing real URLs, passwords, or API keys.

Remote hands for coding agents: an MCP server that runs OpenCode / Claude Code / Codex work on a remote host over SSH, keeps credentials and the local machine closed, and audits everything

Autonomous B2B brand growth agent that sources new media creators & orchestrates partnership campaigns (Astute style)

Local-first, model-neutral memory for coding agents. One shared, owner-governed store across OpenCode, Claude Code, Codex, and Cursor, with optional hosted sync through AI Passport.

Make any operation safe to retry. Crash-safe idempotency for HTTP requests, queue messages, and function calls, in Python and TypeScript. Redis, Postgres, MongoDB, DynamoDB.

How do you KNOW if a 'skill' is any good - deterministically? This runs controlled skill-vs-no-skill experiments. It doesn't just ask, “Did the skill work?” It asks “Did we actually give the skill to the model, did the comparison remain fair, what changed, what did it cost, and do we have enough evidence to say that confidently?”

🛞 scout; a Strands agent that sees, thinks, talks & drives a FrodoBots Earth Rover Mini+. Vision · language · action. ECoT dataset on 🤗 cagataydev/scout-earthrover-ecot

把模糊想法整理成可直接执行的 Agent 提示词：自动选角、逐轮澄清、权限边界、验收标准与验证证据。

Docker image running OpenCode (AI coding agent) and code-server (VS Code) in a single container, with Traefik integration, mise toolchains, and SSH support -- designed for self-hosted NAS/VPS use.

Reusable TypeScript engineering skill for AI coding assistants. Practical guidance on type safety, runtime validation, architecture, security, and testing. Works with Codex, Claude Code, Copilot, Cursor, and more.

GenPark AI Agent Skill - Conversational audio acoustic filler injector masking LLM inference and TTS latency with contextual conversational cues.

File-native Brain v8.3 for Claude Code, Codex, and ZCode — persistent identity, structured memory, cognitive loops, and local recall.

Cursor / Agent Skill to chat with Grok Bot teammates and create new Grok Bots from the CLI

A runnable Docker Compose lab demonstrating the three structural failures of observe-and-detect agent governance. No API keys, no cloud accounts, 60 seconds to run.

Jev (TypeSafe System One) inner loop for ego-browser — one ~0.4s typed decision per DOM step instead of an LLM turn. Agent skill for ego lite.

Agent Skills for OpenKnowledge — install via skills.sh

Compiler-verified multi-agent AI that gets smarter from its own failures. Local-first, AGPLv3, free and open source.

Independent evidence-gated second brain for understanding and porting DeepSeek Harness architecture.

Read-only-by-default MCP server for one or many SQL Servers: schema discovery, SELECT-only queries, execution-plan analysis, opt-in writes.

Valhalla is a peer-to-peer communications runtime for agents. Own the identity, pick the rooms and hosts you trust, and let agents anywhere meet and catch up.

Your agent draws the architecture. Editable Draw.io and Excalidraw diagrams - cloud architectures, system designs, data pipelines and flows - for Claude Code, Codex, Cursor, OpenCode, Copilot, Antigravity and Pi.

Opensource Equity Research & Market Intelligence Agents 🌟 Star if you like it!🌟

Your AI, haunting your browser. A Chrome extension and a tiny CLI that give any MCP client the Chrome you already use.

AI-driven, CLI-first dotfiles for Linux & macOS — chezmoi + mise, Fish, Neovim, and a typed Python CLI

ABAP development from Claude and other MCP hosts: read and write source, transports, unit tests, ATC, dumps, debugger. Multi-system, with server-side guard rails.

Microsoft Foundry Portal에서 직접 해보는 Evaluation Engineering

Tested accessibility skills, CLI, MCP server, and GitHub Action for agent-assisted audits and regression checks.

Infinite context without the infinite context window. Compresses agent sessions (Claude Code, Codex, Grok, Hermes) into a searchable HKP1 pack — full-history RAG, ~60x lower context cost, 0 deps.

Trojan Hippo: Weaponizing Agent Memory for Data Exfiltration — benchmark, adaptive red-teaming, and defenses for persistent memory attacks on LLM agents (arXiv:2605.01970)

Retired. Superseded by Excalibur (xcb), the local workspace for coding agents. Historical OOMPA source and release contracts are preserved here.

V4 learner template: Astro personal branding site + multi-agent labs (Use this template)

An agent ready Node.js Utility to open urls in browser & more

An embedded browser for any AI tool — open ChatGPT, Claude, Gemini, Grok, DeepSeek and more without leaving Obsidian. Send vault notes as context, save AI responses to your vault, and switch between services instantly from the sidebar.

증권사 애널리스트가 만든 한국 상장사 ESG 데이터 MCP 서버 — 5개 기관 ESG 등급·지속가능경영보고서·기업지배구조 지표·온실가스 배출량을 Claude·ChatGPT 등 AI 에이전트에서 바로. API 키 불필요.

Zero-dependency heuristic adversarial prompt injection & override neutralizer (Lakera / NeMo)

What happens when jev meets jq? Intelligence you can pipe for quick experimentation and scripts

A knowledge graph engine for codebases that : recorded intent, co-change context, ranked blast radius, architecture context, precise code discovery, continuous history, visible context accounting and freshness awareness.

黑马程序员Python入门学习与项目练习合集，涵盖基础语法、文件处理、网络爬虫、数据分析与 AI 应用开发

Autonomous AI agent to discover, score, and match job opportunities based on your profile and skills.

Integrate the powerful PI coding agent directly into VS Code with a user-friendly GUI.

Code for "When Agents Slow Down: Understanding LLM Agents' Test-Time Strategies via Elo-per-token Analysis" (arXiv:2609.15309) — paper: https://arxiv.org/abs/2609.15309

Multi-agent framework for Claude Code, Codex & OpenCode — coding agents you can watch, message and hold to rules

RavenClaude — private Claude Code plugin marketplace: governed multi-agent teams, Thing tribunal, comfort-posture dashboard, Runes ready-queue, Prompt Builder, plugin lifecycle, and 184 plugins. Level up from one Claude to a full AI engineering org.

Autonomous hierarchical multi-agent team (CEO -> Manager -> Workers) for Model Context Protocol (MCP). Mix Claude Code, Antigravity, Codex & Direct API with persistent SQLite queues and a real-time web UI.

Local-first, battle-tested, memory that never disappears. Hybrid vector search + knowledge graph + full-text retrieval for AI agents.

ZYRAXON AI — The World's Most Advanced Desktop AI Agent.

基于 DeepSeek Harness Desktop 的 AI 安全测试工作台：支持常规测试、Nday 发现与 0Day 挖掘，整合工具、MCP 和漏洞情报，提供可自定义的聊天协作、子代理调度与证据整理。

面向 Codex、Claude Code、Cursor、Kimi 等 AI Agent 的本地 Xshell MCP 安全桥接：读取终端、合并步骤确认命令、安全上传/下载文件，密码由用户亲自输入，并硬性拦截删除、格式化、数据库清空等危险操作。适用于服务器部署、巡检、排障和离线运维。

Automation toolkit for the App Store Connect REST API: pull catalog metadata, mutate fields (dry-run by default), attach builds, submit versions for review. Generalized; no real ASC credentials or app data. Apache 2.0.

Agent skill that makes mobile UI look like a shipped app: rules with real numbers, screen recipes and a fast screenshot review, from 2,600 screens of 44 iOS apps. Claude Code, Codex, Cursor, Gemini CLI. By Achref Arabi.

Autodesk 3ds Max MCP adapter for AI agents to inspect and edit scenes, automate modeling and animation, and use verified undo through the DCC-MCP ecosystem.

Headless MCP server that gives AI agents the full power of JADX for Android (APK/DEX/AAB) reverse engineering.

Jev agent authorization for MCP tool calls: Kinde identity and permissions plus Jev's typed, calibrated decisions, checked server-side before every call runs

Kavach: fail-closed runtime governance kernel for autonomous AI coding agents. Lifecycle gates, mistake ledger, 3-witness verification. Rust.

Self-hosted AI memory for people and teams: project continuity, traceable evidence, private workspaces, and MCP integration.

Your plan, watching itself. Local desktop roadmap tree that Claude Code and any MCP host keep up to date live. Plain JSON, no cloud.

Reasoning foundation for trustworthy agents.

AI 长篇网文写作工坊：用 AI 帮你从灵感写到成书——分层记忆抗失忆、创意工坊预设市场、未发表稿件零上传、可选本地去 AI 味自检。

AstrBot QQ 音乐插件，支持扫码登录、多账号、刷听歌时长与年/月/周报查询

Give coding agents trustworthy, cited answers about your codebase. Code graph + MCP server, no LLM required.

SOAT — open-source infrastructure for production-ready AI agents.

AI plugins and integrations that enable SAP LeanIX customers to interact with SAP LeanIX through natural language using MCP (Model Context Protocol).

GoodMem Claude Code plugin — Skills + MCP server for memory infrastructure

Index of Matt Busel's Rust crates and tools for LLM agents and infrastructure (memory, budgets, CRDT sync, WASM agents, LLM clients, orchestration, cost dashboard) plus market-data crates, with install lines.

AI agent safety layer for crypto transactions - blocks honeypots, blacklists, and rug pulls before they happen

Agent Ontology Kit — a portable skill that makes AI agents understand a business before they act. Extracts four-layer ontologies (upper/domain/task/application) from companies, APIs, markets, and codebases.

Spider is an independent VS Code extension that provides an AI coding-agent experience inside VS Code using the official Cursor TypeScript SDK, Ollama Local models and OpenRouter.

The first MCP server whose actions burn a token. Let any AI agent price, hold, and burn $THREE on-chain — every burn splits to the incinerator + treasury via the live three.ws token rail, with Jupiter USD pricing. Deflation as an agent primitive.

MCP server for IBM watsonx.ai by three.ws — chat, text generation, embeddings, tokenization, and model discovery with IBM Granite foundation models. Uses your own IBM Cloud credentials.

An AI agent's own autonomous-execution control plane over MCP — set autopilot scopes, a daily SOL spend cap, and auto-execute, then run the propose → execute → undo loop. The agent spends SOL and never sells $THREE. Authenticated, write-heavy; scope + spend enforcement is server-side.

Give an AI agent a real Solana wallet without giving it your money. The agent gets a spending allowance the SPL Token program enforces on-chain, plus guardrails (per-tx cap, daily cap, recipient and host allowlists, expiry, kill switch), x402 payments, and a one-call on-chain deploy. Custody never leaves your wallet: you can revoke or withdraw at a

Resolve Solana .sol names and check *.threews.sol agent-identity availability from any AI agent. sns_resolve maps a .sol name to its owner wallet, sns_reverse maps a wallet to its primary name, and threews_availability checks whether a three.ws agent handle is free to claim. No key, no signer, no payment.

🐾AI browser agent extension with a full LLM agent harness running inside Chrome/Edge. CDP automation, vision, BYOK

面向产品经理的 AI 工作流：11 个 skill，把需求写成完整 PRD，再做成高保真可交互原型