Custom agents

Exam AB-100: Agentic AI Business Solutions Architect

The first open-source GEO linter. 92 rules for SEO, GEO, and content quality — built for AI agents to run, read, fix, and re-lint automatically.

Turns issues into merged PRs/MRs with zero human intervention — supports GitHub and GitLab, and Claude Code, Codex CLI, Kiro CLI, Antigravitiy CLI, opencode, and other agent CLIs.

Turn any MCP server into a CLI in one command.

Autonomous fiction engine — three agents publish a fictional Wikipedia. Hermes-4-70B (canon + critics) + Kimi-K2.6 (prose) + FLUX + ElevenLabs. Constructed-language phonological drift across eras.

Zihuan is a high-performance, highly customizable Rust-based AI Agent platform. Zihuan Impls a interesting Workspace Agent (zihuan code for coding) and an IM Agent (QQ bots, etc.)

🤖 Визуальный конструктор Telegram ботов с drag-and-drop интерфейсом. Создавайте ботов без программирования, получайте готовый Python код для aiogram. React + TypeScript frontend, Express.js backend, PostgreSQL. Поддержка медиа, клавиатур, условной логики и аналитики пользователей.

PileaX is an all-in-one AI knowledge base system. 🍀

Research and product workspace for Hussh personal AI experiences, consent flows, and applied agent systems.

Claude Code skill adapting DeepTutor's graph-enhanced RAG tutoring methodology. Knowledge graphs, dual-loop solving, research reports, adaptive questions. Based on HKUDS/DeepTutor.

Extract equations and context from research papers for LLM coding assistants (arXiv, LaTeX, RAG)

An AI-powered DJ mixing system that crafts custom mixes from your local MP3s using natural language prompts, analyzing BPM, key, and energy for seamless transitions like crossfade and EQ sweep. Ideal for personal use, it generates a mixed MP3 and detailed JSON setlist/planning files.

Open-source AI social-media workflow engine for planning, drafting, brand voice, image generation, and platform-aware content automation.

Chinese traditional fortune-telling AI skill (BaZi 八字 + Zi Wei Dou Shu 紫微斗数) for Claude Code, Codex, Cursor, Trae and any agent platform. PolyForm Noncommercial.

A skill graph for design engineering: animation, transitions, taste, typography, color, components, anti-patterns.

GitHub Actions bot that automatically reviews pull requests using the Claude API, posting inline comments and a summary on every PR.

AI Agent trend judgment with multi-observer sources, emerging-project discovery, scoring rationale, weekly decisions, and tool fallback.

Open-source agentic workspace for professional knowledge work with project memory, client memory, skills, and knowledge workflows.

AI agent firewall that intercepts tool calls (file, shell, network) and enforces deterministic policies at sub-microsecond latency using CEL, IFC, secret scanning, and audit logging.

全球酒店实时推荐及闭环预订Skill，200 万+酒店，全球第三大酒旅B2B公司数据源，免费无调用上限。可按地点、日期、星级、人数、标签等条件搜索酒店列表，返回候选酒店及展示价。可查询指定酒店的实时房型列表、价格计划和取消政策。可锁定价格，创建订单+支付宝二维码，并查询历史订单。

Zero-dependency CLI to install AI agent skills directly from any source. No marketplace, no registry, no signup — just point it at a local folder or a GitHub repo and it works. Works with opencode, claude-code, cursor, and all spec-compliant agents.

Local-first AI contract review agent that turns legal templates and counterparty drafts into clause-level analysis and tracked-change DOCX redlines.

Provide different cloud products MCP Server tools to help developers manage cloud resources with AI-agent

Praxen — agent behavior verifier. Compares an AI agent's declared policy against the available evidence; reports where observed behavior diverges from declared intent.

Harness Study · The Engineering Practice for AI Agents — a from-scratch tutorial on agent harness engineering: 8 runtime mechanisms, a cross-cutting Safety control plane, verifiers, trajectories, and self-evolution. Bilingual (English/Chinese).

Self-installing personal AI orchestrator. Hand the latest blueprint file to a Claude Code session and it builds a full multi-agent system. Telegram interface, persistent memory, specialist crew.

Social leaderboard for LLM token usage — sync your agent usage with the CLI, climb the leaderboard

一款集成了ReAct，Plan-and-Execute和Multi-Agent的coding agent

An intelligent traffic monitoring system that collects traffic flow and metrics including average speed and vehicle counts for each road. Features real-time data visualization with interactive dashboards and an integrated AI Agent chatbot for querying processed traffic data in real-time.

A curated skillstack for AI agents. Make your AI code editor write code like a Tech Lead. Stop copy-pasting instructions. Install 794 expert rules, workflows, and custom personas in Cursor, Claude Code, and Windsurf with one CLI command.

AI agent harness tool allowing it to wipe its own session context and re-run the first message

AI-native factory for Go + SolidJS apps — Cursor rules, spec drift CI, plans, and a production starter with auth, SSR, SSE, Cloud Run deploy.

The complete backend for AI agents in one SQLite file : Documents, vectors, cache, queue, and cron.

Talk to your Claude Code by voice — a real voice call with your terminal agent. Local whisper STT, edge-tts voice, no API key.

An LLM-powered social simulation and roleplaying engine where NPCs have their own lives too.

Task-plan driven SEO and LLM-friendly website skill cluster

AI agent skill for TR Dizin (trdizin.gov.tr) — search Turkish academic publications, journals, authors & institutions via the open JSON API, with advanced field search, references/citations, and PDF-to-text. No browser, login, or API key; works with Claude Code, Codex, and other agents.

[ICML 2026] Official implementation for paper "Unsafer in Many Turns: Benchmarking and Defending Multi-Turn Safety Risks in Tool-Using Agents"

The AI-native technical SEO crawler. Open-source MCP server for Claude / Cursor / Codex — 37 tools, 50+ checks, unlimited pages, WAF detection, ephemeral by design. Built on LibreCrawl. MIT.

AI-native publishing system — build books like software. 6-phase pipeline from blank page to EPUB/PDF, with 16 skills, 7 agents, and 8 genre templates. Claude Code plugin.

Explore the ecosystem of open source tools, agents, runtimes, platforms, and infrastructure for the AI Native era, curated by hand and evaluated across multiple dimensions.

Android SDK and MCP server: 157 tools for calendar, contacts, SMS, files, camera, sensors, notifications and UI automation, for on-device LLMs and MCP clients such as Claude Code and Cursor.

An open-source AI assistant for work and everyday life. Remembers context and helps you get things done. Local-first, on a computer you choose.

Local dashboard for Claude Code, Codex, OpenCode, OpenClaw and Qoder session history, token usage, cost tracking, tool calls and subagent replay.

Proof-driven Claude/Codex agent loop with paired execution, persistent resume, and adversarial security/performance gates.

Hermes-first operational AI substrate with filesystem-grounded memory, governance, persistent routes, and recoverable agent continuity.

Cairn v0.6.39 is a cross-platform desktop Git client combining repository administration, terminal-based work, and AI coding-agent workflows in one application for developers managing Git projects.

Give Claude Code a persistent memory — it remembers you, your projects, and your decisions as plain Markdown files you own. Fully local. Just say `start`.

A disciplined, delegation-driven operating system for AI coding agents. npx contractor-kit.

Reflective memory for agents

[ICLR2026] DriveAgent-R1: Advancing VLM-based Autonomous Driving with Active Perception and Hybrid Thinking

Autonomous Meteora DLMM liquidity-provider agent on Solana — single headless Rust binary, Telegram-controlled

Swift source, Apple guides and MCP tools for AI coding agents. Review iOS code, scaffold apps, and build/run/preview with Xcode Simulator.

Production local AI and offline LLM app for iPhone, iPad, Mac, and Apple Vision Pro—available on the App Store with open-source Swift code.

Complexity-aware agentic coding pipeline for Claude Code

BBARIT Terminal (빠릿터미널) — free AI vibe coding IDE, terminal & agent in one, built 100% in Rust. Ships ⚡ BBARIT Agent (built-in coding agent with a verify loop) + a 🦌 dev↔review AI pair, runs Claude, Codex, Gemini, Kimi, Qwen & Ollama side by side, and edits Word/Excel/PowerPoint. macOS & Windows. Free.

Autonomous agentic system for supervised machine learning model development.

AI video auto-dubbing plugin for coding agents — translate & dub videos into multiple languages via the Perso Dubbing API. Works with Claude Code, Cursor, Codex & Antigravity.

[NeurIPS 2026] CoffeeBench: Benchmarking Long-Horizon LLM Agents in Heterogeneous Multi-Agent Economies

Autonomous research loop skill for LLM agents — inspired by Karpathy's autoresearch. Define a goal, let the agent iterate.

Agent-skills marketplace for Coddy, Claude Code, Codex and Cursor: RPA BDD workflow, Logika (Chelpanov formal logic), token-cost, MikroTik config generator

OpenTelemetry integration for AI Agents Cursor / Copilot / Gemini / Claude / Codex that records agent activity and exports detailed telemetry with OTLP protocol

CADSmith (formerly AutoFab) is a multi-agent system that generates manufacturing-ready CAD models from natural language by combining LLM code generation with programmatic geometric validation from the OpenCASCADE kernel and visual inspection from an independent vision-language model, iteratively refining until the part is dimensionally correct.

Workspace for documents with AI

37 folder-drop skills for Microsoft 365 Copilot Cowork. 21 workflow skills plus a 16-persona executive review board. Drop a folder into OneDrive, no admin, no app package. The same SKILL.md files run verbatim in Claude Code.

面向 AI Agent 的 HarmonyOS / ArkUI 设计与动效 Skill、规则库与评测基线 · Design & motion Skills, rule library and evaluation baseline for HarmonyOS / OpenHarmony / ArkUI

Fast, pure-Python full-text indexing and search. Actively maintained continuation of Whoosh / whoosh-reloaded. Install: pip install whoosh3

基于 Go 与 React 的视频知识库与 Agent 问答平台，支持多模态检索、跨视频分析与来源回放

Self-hostable, MCP-native testing platform. Manual test management, deterministic runs, optional AI. Your stack, your LLM, your data.

Declarative automation for repetitive steps in herdr

Compact AI agent runtime control plane with governed tools, HITL approval, resumable execution, trace-driven evaluation, and SWE-bench-shaped evidence.

The Cost of Remembering: filesystem memory matches long-context accuracy on LongMemEval while reading 97% fewer tokens and costing 95% less. Harness, run data, 129 agent-built memories, and paper source.

Measured Agent Skills for geospatial AI — 18 skills that stop when the data cannot support the claim. Routing benchmark, held-out split, and known defects published.

An open-source, local-friendly runtime for agentic AI companions — an always-on brain with self-editing memory. You own the model, the memory, and the character.

AI agent operated by @oliver-zehentleitner

A beginner-friendly, notebook-first educational project: building LLM Agent Execution Harnesses from scratch with runtime observability. / 面向零基础的 Notebook 教学向项目：从零构建大模型 Agent Execution Harnesses (执行基础设施)与运行时可观测性。

通用视觉识别 Skill 模板：为无多模态能力的 AI agent（如 DeepSeek）补上图片识别能力，支持本地图片与 URL。纯净模板 + scripts/setup.py 一键配置 Key/端点/模型，遵循开放 Agent Skills 标准，兼容 Claude Code / Codex / OpenClaw / Reasonix。

AI-powered developer command center for GitHub context, progress narratives, smart TODOs, project notes, hackathon planning, and a context-aware copilot.

Claude Agent SDK + LLM-wiki，就是最强大的agentic RAG 。让 AI 当一个“知识编译器”，先把你的所有资料读一遍，整理成一本结构清晰、彼此关联的百科全书。以后你再提问，AI 直接翻这本百科书就行了。 本项目主要展示了我制作的 llm-wiki 怎样通过 Claude Agent SDK 转为 agentic rag， 效果非常好的。使用方法：.venv\Scripts\Activate.ps1; python 7_wiki_writer.py --request "怎样拼搭微型车"

給 Claude Code、Codex 等 agent 用的 macOS 磁碟清理 skill：診斷「系統資料」佔用、唯讀掃描腳本、以及 10 個會造成誤判或資料損失的陷阱

AntNest 是一个采用蚁后 / 工蚁架构（或蜂后 / 工蜂）的 AI 编程助手。你可以把它想象成一个蚁群： 蚁后（本体）：只负责思考、规划、决策。蚁后从不亲自执行任何命令。 工蚁（复制体）：蚁后需要干活时，会生成一只工蚁——把自身代码复制一份，派到隔离的临时目录里执行命令。工蚁干完活把结果交回来，然后被立即销毁。

Open-source desktop AI agent with an embedded browser runtime and human-in-the-loop control.

Google Search Console MCP Server for Claude, Cursor, Windsurf and other MCP Clients

Cross-platform SSH and SFTP client with organized server assets, terminal workflows, port forwarding, secure sync, and AI assistance.

Desktop cockpit for DeepSeek Harness (dsh): token usage & cost tracking, budget alerts, runtime auto-update with rollback, Quick Ask hotkey, scheduled tasks, session search. Win+macOS. DeepSeek Harness 桌面驾驶舱：成本/用量监控 · 自动更新 · 定时任务

Your agent loses the plan between sessions, so saga-mcp gives it a SQLite-backed tracker for projects, epics, tasks, subtasks, dependencies, and notes: 41 MCP tools, an activity log, and a one-call dashboard, with no accounts and no external services.

Korgo Bot: persistent AI bots with built-in Orgo cloud computer use.

Local-first desktop command center for Claude Code, Codex, OpenCode, and other coding agents—parallel tasks, worktrees, dev services, port forwards, diffs, and review.

Unofficial MCP server for T-Bank (Т-Банк, ex-Tinkoff) mobile banking: accounts, cards, transfers, bill pay, grocery ordering, tickets & travel — for Claude Code and any MCP client.

The open-source cloud for the agentic era. Don't pick a database, a region, or a RAM slider - your coding agent and Bay decide, then keep the app alive and scaling 24/7

20-pass adversarial audit for vibe-coded software. Injection, IDOR, races, N+1, idempotency, transactions, memory leaks — then a false-positive verifier.

RAG-powered document intelligence assistant built for the Midnight Hackathon AI Track. Users upload confidential documents (medical records, contracts, financial statements, legal files) and interrogate them through a natural-language chat interface. The system returns grounded, cited answers without exposing raw document contents unnecessarily.

Connect Claude to your Zotero library — Zotero plugin that mirrors collections as agent-readable folders with Markdown, BibTeX, and AI reviews.

Dev Boards + BeeZilla — mission kanban for AI agents and a consumer orchestrator, on one engine. MIT, self-hostable.

Drive OriginLab Origin scientific plotting from AI chat via MCP. 74 tools: modal-dialog watchdog, LabTalk safety gate, FigureSpec declarative YAML, fit params, matrix tools, matplotlib bridge, PPT assembly, editable OPJU delivery. Windows + Origin 2021+.

Declarative Nix configuration framework for the Pi coding agent

Java Web 全链路自动化安全审计系统。基于 Claude Code Agent Teams，一行命令完成路由识别→调用链追踪→8种漏洞检测→利用链编排→审计报告。覆盖 SQL 注入/反序列化/SSRF/表达式注入/XXE/文件上传/文件读取/鉴权绕过 + OSV.dev 3000+ CVE。支持快速模式、增量审计、CFR 反编译。

AI Agent Skill for B2B lead research, decision-maker discovery, customs checks, outreach drafting & controlled sending. Works with Claude Code, Cursor, Cline & Hermes.外贸客户挖掘 AI 技能包：找海外买家、挖决策人、查海关信号、写开发信、受控发送。支持 Claude Code / Cursor / Cline / Hermes 四平台。

Click does not replace Nx, Bazel, pytest, or Jest. It sits above existing verification tools and manages which verification evidence remains applicable across AI-agent iterations.