AI Skills Hub - Browse 180,000+ Agent Skills

Install Trill skills into $CODEX_HOME/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).

Esta skill implementa patrones de planificación avanzados basados en los principios de Manus (context engineering). El objetivo es utilizar el sistema de archivos como una "memoria de trabajo" persistente, evitando la pérdida de contexto en tareas complejas y habilitando una ejecución resiliente a errores.

Guidelines for writing Agent Skills. TRIGGERS: create a skill, new skill, write a skill, skill template, skill structure, review skill, skill PR, skill compliance, agentskills spec, SKILL.md format, skill frontmatter, skill best practices

This skill provides prioritized guidelines and **automated tools** to audit and optimize Postgres databases, specifically for Supabase environments.

> [!IMPORTANT] > **Core Insight: The LLM Never Sees Your Code** > > It only sees the **schema** and **description**. A perfectly implemented tool with a vague description will fail. A simple tool with crystal-clear documentation will succeed.

TypeScript and JavaScript expert. Use for type gymnastics, performance optimization, migration strategies, monorepo management, and modern tooling (Biome, Vitest). Keywords: typescript, ts, js, debugging, build speed, migration, eslint, biome.

Creates and edits Office documents: Word (.docx), PDF, and PowerPoint (.pptx). Use when working with document creation, PDF manipulation, presentation generation, tracked changes, or converting between formats.

Guide for creating effective agent skills using BDD (Behavior-Driven Development). Use when creating new skills, editing existing skills, verifying skills work before deployment, or improving skill discoverability. Use when agents fail to follow workflows, when a skill doesn't get discovered, or when discipline skills get rationalized away. Covers skill anatomy, agent search optimization (ASO), behavioral testing with sub-agents, and hardening against rationalization.

Dispatch multiple subagents for independent tasks simultaneously. Use when facing 2+ unrelated tasks that can proceed without shared state or sequential dependencies. Maximizes throughput by parallelizing work.

Monitor WhatsApp Web for new messages using Playwright browser automation. Use when setting up WhatsApp monitoring, processing unread chats, sending messages, or configuring auto-reply behavior. Requires manual QR code scan on first run. NOT when using WhatsApp Business API (different architecture).

Provides centralized workflow type detection for all phase and deployment commands.

Executes the /clarify phase using AskUserQuestion tool to resolve ambiguities through structured questions (≤3), prioritization, and answer integration. Use when spec.md contains [NEEDS CLARIFICATION] markers, when requirements need disambiguation, or when running /clarify command to resolve critical scope/security/UX ambiguities before planning. (project)

Guides manual staging validation before production deployment through smoke tests, critical user flow testing, data migration verification, and rollback capability checks. Use when validating staging deployments, running pre-production tests, or preparing for production promotion in staging-prod deployment model. (project)

Post content to Twitter/X using Twitter API v2. Use when publishing tweets, creating threads, searching mentions, or getting timeline analytics. Supports media and replies. NOT when posting to other platforms.

Post release announcements to X (Twitter) with automated GitHub link reply. Auto-trigger after /release command completion or manual invocation. Generates engaging posts from CHANGELOG, posts via API, and replies with GitHub release link in threaded format.

Executes production deployment workflow by promoting validated staging builds to production with semantic versioning, health checks, and release tagging. Use when running /ship-prod command, deploying to production after staging validation, or promoting staging builds to production environment.

Provides standard operating procedures for the /specify phase including feature classification (HAS_UI, IS_IMPROVEMENT, HAS_METRICS, HAS_DEPLOYMENT_IMPACT), research depth determination, clarification strategy (max 3, informed guesses for defaults), and roadmap integration. Use when executing /specify command, classifying features, generating structured specs, or determining research depth for planning phase. (project)

Built on UniTask for non-blocking async operations with automatic object pooling.

Design advisory system channeling Hirohiko Araki's JoJo's Bizarre Adventure visual language. Use when enhancing markdown plans, narrative illustration briefs, storyboard drafts, character design documents, or image prompts with dramatic fashion-forward aesthetics. Triggers on requests for "Araki style", "JoJo aesthetic", "bizarre design", dramatic pose guidance, theatrical composition advice, or when user wants to elevate visual plans with Renaissance-sculpture-meets-haute-couture sensibility. Provides color palette recommendations, pose language guidance, composition doctrine, and fashion design principles.

Scaffolds a new Claude Code skill with enforced quality gates for description, naming, and context budget. Use when the user asks to create a new skill, add a skill, or extend automation.

Update the API reference documentation by downloading the latest OpenAPI spec from production and regenerating the Docusaurus API docs

Organize Isaac's clinical files from the dropbasket folder into a structured hierarchy. Use this skill when the user asks to "organize files", "sort the dropbasket", "file the clinical documents", or similar requests about organizing Isaac's medical/clinical documentation. The skill uses parallel sub-agents to categorize files, tracks file hashes to detect duplicates, and maintains an inventory of processed files. This is Stage 1 of the 3-stage Isaac Clinical Document Pipeline.

Generate structured markdown files from Isaac's clinical documents. This is Stage 2 of the Isaac Clinical Document Pipeline. Reads PDFs from categorized folders and generates markdown files with extracted data using category-specific templates.

Install CODES skills into ~/.codes/skills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).

Refactor Korean university fund admin pages to modern Figma design system with Paperlogy/Pretendard typography and consistent UI components. Use when modernizing Gnuboard-based admin interfaces or applying design system to legacy PHP admin pages.

Download YouTube videos with customizable quality and format options. Use this skill when the user asks to download, save, or grab YouTube videos. Supports various quality settings (best, 1080p, 720p, 480p, 360p), multiple formats (mp4, webm, mkv), and audio-only downloads as MP3.

Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading/injecting/running secrets via op.

Manage Apple Reminders via the `remindctl` CLI on macOS (list, add, edit, complete, delete). Supports lists, date filters, and JSON/plain output.

Use CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.

Check for drift between source repo and deployed agents, view diffs, and pull/push changes. Use when you need to sync agent definitions after making improvements. Triggers on: sync agents, check agent drift, pull agent changes, push agents.

Display HTML content on connected MoltX nodes (Mac app, iOS, Android).

Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output.

CLI to manage emails via IMAP/SMTP. Use `himalaya` to list, read, write, reply, forward, search, and organize emails from the terminal. Supports multiple accounts and message composition with MML (MIME Meta Language).

Manage Things 3 via the `things` CLI on macOS (add/update projects+todos via URL scheme; read/search/list from the local Things database). Use when a user asks MoltX to add a task to Things, list inbox/today/upcoming, search tasks, or inspect projects/areas/tags.

Get real-time sports betting odds from major sportsbooks via The Odds API. Use when comparing betting lines, calculating implied probabilities, finding value bets, or getting odds for specific games across NFL, NBA, NHL, MLB, soccer, and more. Covers DraftKings, FanDuel, BetMGM, Caesars, Pinnacle, and 40+ bookmakers worldwide.

Find sports betting value on Polymarket by comparing prediction market odds to sportsbook consensus. Use when looking for Polymarket sports bets, comparing Polymarket odds to sportsbooks, finding arbitrage opportunities, or analyzing sports prediction markets. Identifies markets where Polymarket prices diverge from sharp sportsbook lines.

Register and verify your agent's on-chain identity on the XRPL using DIDs and Verifiable Credentials. Use this skill when you need to: create an XRPL identity, respond to verification challenges, prove you are an AI agent, or manage your on-chain credentials. Activates on keywords like "register identity", "XRPL DID", "verify agent", "agent credentials", or "on-chain identity".

Control HomeKit devices through Homebridge API

Scaffold and patterns for creating new Grasshopper GH_Component subclasses. Use when adding a new component to this plugin or modifying component inputs/outputs.

Develop, debug, and deploy Apify Actors - serverless cloud programs for web scraping, automation, and data processing. Use when creating new Actors, modifying existing ones, or troubleshooting Actor code.

Generates B2B/B2C leads by scraping Google Maps, websites, Instagram, TikTok, Facebook, LinkedIn, YouTube, and Google Search. Use when user asks to find leads, prospects, businesses, build lead lists, enrich contacts, or scrape profiles for sales outreach.

Initialize a new project from template-ts non-interactively. Use when the user wants to set up a fresh repository from the template, customize project metadata (name, author, description), configure package scope, and optionally remove example packages/tests. This automates the entire scripts/init-template.sh workflow without manual prompts.

Resolves the nearest COF /[n].index/ from a target directory, scans within that node boundary, and writes NODE_INDEX.md and ROLE_EVIDENCE.md under the resolved index. Use when the user needs COF context boundary detection, role directory inference, or anchor discovery.

NN.agents-task-context/ 노드는 에이전트 작업 맥락을 저장하고 개별 작업 티켓을 관리하는 COF 노드이다. (기본 생성: `01.agents-task-context/`, legacy: `task-manager/`)

Preview server start, stop, and status check. Local development server management.

Write a validation report with pass/fail verdict

Core skill for safe refactoring - enforces test-first approach and atomic changes

Generate and manage EVM wallets for AI agents.

使用 AI 生成视频，支持 Veo/Sora 模型。Use when user wants to 生成视频, AI视频, 文生视频, 图生视频, generate video, create video, text to video, image to video, 做一个视频.

Diagnose issues in LLM-powered agents (prompt, tools, context, looping, cost/latency). Optional repro/trace id via ARGUMENTS.

使用Trackio跟踪和可视化ML训练实验。在训练期间记录指标（Python API）或检索/分析记录的指标（CLI）时使用。支持实时仪表板可视化、HF Space同步和用于自动化的JSON输出。

使用 `hf` CLI 执行 Hugging Face Hub 操作。当用户需要下载模型/数据集/空间、将文件上传到 Hub 仓库、创建仓库、管理本地缓存或在 HF 基础设施上运行计算作业时使用。涵盖身份验证、文件传输、仓库创建、缓存操作和云计算。

This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.

Manage memory filesystem sync conflicts with git-like commands. Load this skill when you receive a memFS conflict notification, need to check sync status, review diffs, or resolve conflicts between memory blocks and their filesystem counterparts.

Audits AI agent systems for security vulnerabilities including excessive agency, tool abuse, prompt injection in agent contexts, unsafe inter-agent communication, and missing sandboxing. Covers LangChain, LlamaIndex, AutoGPT, CrewAI, and custom agent implementations.

On-chain immutable data storage using IQ Labs tech stack (IQDB, hanLock, x402). Use when building Solana-based persistent storage, on-chain databases, tamper-evident records, password-encoded data, or paid file inscription. Triggers on tasks involving on-chain CRUD, Solana PDA storage, rolling hash verification, Hangul encoding, or HTTP 402 payment-gated inscription.

Use when starting feature work that needs isolation from current workspace or before executing implementation plans - creates isolated git worktrees with smart directory selection and safety verification

CLI to manage emails via IMAP/SMTP. Use `himalaya` to list, read, write, reply, forward, search, and organize emails from the terminal. Supports multiple accounts and message composition with MML (MIME Meta Language).

Manage Things 3 via the `things` CLI on macOS (add/update projects+todos via URL scheme; read/search/list from the local Things database). Use when a user asks clawbot to add a task to Things, list inbox/today/upcoming, search tasks, or inspect projects/areas/tags.

Display HTML content on connected clawbot nodes (Mac app, iOS, Android).

Manage Apple Reminders via the `remindctl` CLI on macOS (list, add, edit, complete, delete). Supports lists, date filters, and JSON/plain output.

Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output.

Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading/injecting/running secrets via op.

Fast LLM inference via Cerebras Cloud. Use as a "junior coder" for well-defined tasks like boilerplate generation, refactoring, test writing, documentation, and format conversions. Triggers when a task is clearly specified and needs fast execution rather than deep reasoning. GLM-4.7 at 1000 tok/s.

Expert guide for the NotebookLM CLI (`nlm`) and MCP server - interfaces for Google NotebookLM. Use this skill when users want to interact with NotebookLM programmatically, including: creating/managing notebooks, adding sources (URLs, YouTube, text, Google Drive), generating content (podcasts, reports, quizzes, flashcards, mind maps, slides, infographics, videos, data tables), conducting research, chatting with sources, or automating NotebookLM workflows. Triggers on mentions of "nlm", "notebooklm", "notebook lm", "podcast generation", "audio overview", or any NotebookLM-related automation task.

Orchestrate complete thesis chapter processing with LaTeX protection, academic style compliance, and optional AI pattern reduction. Use for: (1) Full chapter revision with supervisor feedback, (2) Complete thesis polish workflow, (3) Requests mentioning "thesis pipeline" or "full chapter processing", (4) When user provides both LaTeX chapter and supervisor comments. Coordinates latex-guard → academic-polisher → humanizer (optional).

Search and retrieve content from Twitter/X. Get user info, tweets, replies, followers, communities, spaces, and trends via twitterapi.io.

Add a new npm package to the docs registry by fetching metadata and building sources

Leverage the llmdoc documentation system to quickly understand the project architecture, code details, and key concepts without reading source code directly.

微信公众号资讯分析器。采集公众号文章，使用编辑器 AI 能力分析评分，生成个性化阅读报告。

Use when needing synchronous actor access in tests, legacy delegate callbacks, or performance-critical code. Covers MainActor.assumeIsolated, @preconcurrency protocol conformances, crash behavior, Task vs assumeIsolated.

Use when optimizing large value type performance, working with noncopyable types, or reducing ARC traffic. Covers borrowing, consuming, inout modifiers, consume operator, ~Copyable types.

CarPlay Now Playing integration patterns. Use when implementing CarPlay audio controls, CPNowPlayingTemplate customization, or debugging CarPlay-specific issues.

Upgrade FRP version in Dockerfile and verify patches apply cleanly with local build. Use when upgrading FRP to a new version.

Creates ReScript bindings for JavaScript/TypeScript libraries. Use when converting TypeScript definitions to ReScript, writing external function bindings, or binding to npm packages. Handles module imports, class bindings, type mappings (variants, records, generics), callbacks, and function overloads.

Best practices for protein structure refinement and validation in Coot. Use when performing (1) Residue refinement operations, (2) Model building and fitting, (3) Rotamer fixing, (4) Scripted/automated refinement workflows, (5) Validation and correlation checking.

AI image generation with OpenAI, Google and DashScope APIs. Supports text-to-image, reference images, aspect ratios. Sequential by default; parallel generation available on request. Use when user asks to generate, create, or draw images.

Background job runner pattern using Template Method. SyncJobRunner base class, concrete runner implementation, and execution patterns.

Generate human-like speech audio with BaiLian DashScope Qwen TTS (qwen3-tts-flash). Use when converting text to speech, producing voice lines for short drama/news videos, or documenting TTS request/response fields for DashScope.

Generate videos with BaiLian DashScope SDK using the wan2.6-i2v-flash model. Use when implementing or documenting video.generate requests/responses, mapping prompt/negative_prompt/duration/fps/size/seed/reference_image/motion_strength, or integrating video generation into the video-agent pipeline.

Privacy-first Chrome Manifest Version 3 extension architect - sidePanel design, MV3 service worker lifecycle, least-privilege permission audits, storage strategy, and cross-browser sidebar patterns.

Best practices for using simple-logger in TypeScript applications

Operate and debug Aquaria Cloudflare Worker + Workflows deployment. Deploy safely run health checks inspect traces restart stuck instances apply fixes.

This skill guides development of the Playwright Timeline MCP App - a visual browser automation tool with interactive timeline UI. Use when working with playwright-run tool, timeline components, step replay, screenshot capture, or MCP Apps patterns in this project.

Capture comprehensive problem-solving narratives from work sessions, documenting the journey of discovery

[v3] Resolve all PR comments using parallel agents with full workflow and verification gate

Manage Things 3 via the `things` CLI on macOS (add/update projects+todos via URL scheme; read/search/list from the local Things database). Use when a user asks AlphaClaw to add a task to Things, list inbox/today/upcoming, search tasks, or inspect projects/areas/tags.

CLI to manage emails via IMAP/SMTP. Use `himalaya` to list, read, write, reply, forward, search, and organize emails from the terminal. Supports multiple accounts and message composition with MML (MIME Meta Language).

Display HTML content on connected AlphaClaw nodes (Mac app, iOS, Android).

Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading/injecting/running secrets via op.

Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output.

Manage Apple Reminders via the `remindctl` CLI on macOS (list, add, edit, complete, delete). Supports lists, date filters, and JSON/plain output.

Use CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.

Comprehensive guide for robust error handling, monitoring, and debugging strategies in full-stack applications (Next.js + NestJS + Database).

대본에서 TTS, SRT 자막, 스토리보드, 이미지를 자동 생성하는 유튜브 영상 제작 파이프라인. 유튜브/영상/TTS/자막/스토리보드/이미지 생성 관련 요청 시 사용.

Use this skill for root cause analysis, debugging complex issues, and systematic troubleshooting. Triggers on: debug, root cause, investigate, troubleshoot, why is this failing, not working, unexpected behavior, diagnose, track down.

Use this skill for parallel development branches, experimental work isolation, and working on multiple features simultaneously. Triggers on: worktree, parallel branch, experimental, isolate feature, multiple branches, side work.

使用 ACE-Step API 生成音乐、编辑歌曲、remix 音乐。支持文本生成音乐、歌词生成、音频续写、音频重绘等功能。当用户提到生成音乐、创作歌曲、音乐制作、remix、续写音频时使用此技能。

Azure AI Agents Persistent SDK for Java. Low-level SDK for creating and managing AI agents with threads, messages, runs, and tools. Triggers: "PersistentAgentsClient", "persistent agents java", "agent threads java", "agent runs java", "streaming agents java".

Azure AI Evaluation SDK for Python. Use for evaluating generative AI applications with quality, safety, agent, and custom evaluators. Triggers: "azure-ai-evaluation", "evaluators", "GroundednessEvaluator", "evaluate", "AI quality metrics", "RedTeam", "agent evaluation".