Self-tailoring skill that builds a personal AI knowledge vault for Claude Code or Codex — INDEX/MAP routing engine, git sync, cross-agent skill-linking. With or without Obsidian. No secrets, no personal data.

Community surf forecasting and board recommendation workflows for AI agents, with private local rider data

Local-first media generation hub and cross-agent MCP/Skill blueprint for Alibaba Cloud Model Studio Token Plan

Claude Agent Skill that strips AI writing tells out of prose. Pattern catalog derived from Wikipedia's WikiProject AI Cleanup.

A lightweight, read-only AI agent skill for reviewing recent code diffs with bounded context, targeted validation, and no automatic fixes. 面向近期代码变更的轻量只读 AI 审查 Skill。

A beginner-first Skill for building your own VPN with Codex, with all expected costs explained upfront and every payment confirmed by the user.

Safely adapt external repositories into an existing codebase through evidence-first analysis, conflict interviews, and verified integration.

Use secrets without revealing them to AI coding agents. Native macOS Keychain, Windows Credential Manager, and Linux Secret Service support.

Find agent skills by reading what they actually do.

Agent Skill: carry one delegated goal from a single sentence to multiple shipped batches, without step-by-step approval. Pressure-tested, tests included.

🎩 Your AI cleans years of bookmark chaos in one conversation. Merges browser profiles, dedupes, repairs dead links, sorts by how you actually browse — backups and one-copy undo at every step.

Turns Claude into your app's worst user: an adversarial client-side chaos-testing skill for Claude Code. Nine personas, a strength dial, and reproducible bug findings.

Official Agent Skills for integrating and auditing LnkFlow

OpenCLI 技能资产与运行时插件适配器的 monorepo（学术检索 + 报销流程）。Monorepo for OpenCLI skill assets and runtime plugin adapters.

🔦探照灯：在 AI 对话中自然触发的领域认知工作流：术语提取 → 七步深度讲解 → 知识卡片固化。支持 opencode / claude-code / cursor。

将员工经验、专家方法和业务需求蒸馏为可安装 Skill 的对话式生成器

Reusable agent skills published through skills.sh

Public agent skills for creating and managing HiGantic artifacts

Claude Code skill that catches the gap between "the file is fixed" and "the running system reflects it" after an AI agent makes a change

AI job-search agent

Publish HTML from your coding agent, share a link anyone can comment on, and pull the feedback back into chat. A skill for GitHub Copilot CLI, Claude Code, and OpenAI Codex.

Manage the skill lifecycle by designing, building, testing, evaluating, and packaging with architecture-first principles and advanced evaluation tools.

Turn websites, Electron apps, and local tools into command-line interfaces using browser sessions and AI-powered discovery.

Generate AI-driven skills efficiently with a comprehensive toolkit designed for creating, managing, and deploying custom AI capabilities.

Equip AI coding agents with a suite of specialist skills to audit, fix, verify, and report on production engineering tasks.

Self-building AI code generation engine that generates application flows instead of implementing them. AGPL-3.0.

A CLI for Transmission BitTorrent client — filtering, seeding policies, tracker management, per-file priority, macOS Keychain auth, and AI agent plugin

graft grafts tree-sitter ASTs

Analyze Sui Move smart contracts with a parallelized tool that checks 143 attack vectors and applies DeFi security best practices.

Forge: skill architect and builder — six-phase pipeline for designing, scoping, building, and validating agent skills.

算盘：智能体时代的个人财务管理CLI工具

Scan AI-related plugins and tools with 13 engines to quickly detect security risks and ensure safe installation before use.

MindForge: The Enterprise Agentic Framework for Claude Code & Antigravity. High-performance autonomous execution, wave-parallelism, and multi-tier governance for production-grade AI engineering.

A CLI to manage your agent skills.

Provide Claude Code with expert guidance to install, configure, and optimize memory-lancedb-pro for long-term AI memory management and retrieval.

Provide financial analysis and trading agent skills designed for integration with Claude Code to enhance decision-making and automation.

Enable AI coding assistants to generate accurate Swift tests using structured testing features tailored to common LLM errors.

Simplify software decisions by applying a skill that favors local, clear solutions before adding complexity or layers.

Enable AI agents to access real-time Danish data for job search, housing, groceries, weather, travel, and health information.

Provide and install pre-packaged plugins to extend Claude Code with new skills directly within your Claude environment.

Claude Code plugin to help write TableTest-style JUnit tests in Java and Kotlin

Round-trip CLI for your personal agent skills library — install, push, detect, fork.

Evidence-backed advisory structure, CLI contracts, and projections for technical advisory workflows.

Offline security and compatibility auditor for AI agent skills

intervals.icu cli

Provide Go agent skills for cleaner, idiomatic code across Claude Code, Cursor, Copilot, and more with one install

Build focused research for Claude Code with one skill, three reference packs, and a clear workflow for mapping, judging, and synthesizing sources

📊 shift-left for skills — track which AI skills get used, know which never get called. Claude Code hook-based.

Build browser games with automated Three.js skills for Codex and Claude Code to manage graphics, logic, assets, and builds.

Convert YouTube channels into AI assistants that answer questions with verbatim quotes and timestamped links.

Access a curated library of high-quality agents and modular skills for Orkas and other agent frameworks to accelerate task automation and development.

rldyour AI CLI configuration for MiMoCode: native .mimocode config, build/plan/compose agents, persistent memory projection, commands, skills, MCP, browser routing, and runtime validation.

Real work, fully auditable, in one file: a standalone, target-agnostic improvement-reasoning skill for LLM agents - examines and improves anything the model can reason about (code, documents, plans, letters) while recording every reasoning step in an auditable trail. Single markdown file, no installer, built on the Principles of Earned Autonomy.

frontend-reference-hunt — a coding-agent skill that finds real visual references (Dribbble, Mobbin, Awwwards, Godly) and maps them into a structured board: register, type, palette, spacing, motion archetypes, steal/avoid notes.

漫画书名一报，剩下的，有请老赵。自动研究、判断、写作并交付网页漫画书评的 Codex Skill。

Evidence-backed, resumable skill ecosystem auditing for the Codex app and CLI.