Skills
101,174 packaged capabilities and playbooks for agents. Install any of them with one command.
All
When the user wants to write YouTube video descriptions, optimize metadata, add timestamps, improve search discoverability, or fix underperforming SEO. Also use when the user says 'write a description,' 'video description,' 'YouTube SEO,' 'tags for my video,' 'timestamps,' 'improve discoverability,' 'my videos aren't showing up in search,' 'metadata optimization,' 'description template.' For title optimization, see title-craft. For end screen strategy, see end-screen-cta.
Clean up UI/UX slop when onboarding any project — scan the existing views, take playwright screenshots (before), let hallmark read the project itself and genuinely REDESIGN it (generating its own design.md, not boxed in by the old slop tokens), then loop on fixes driven by the slop-test 58 gates + `/impeccable audit`/`critique` + `/qc-uiux` (a four-part senior verdict, deterministic `visual-qa` engine) until clean or out of loop budget. Call when the user says "clean the slop", "tidy up the project UI", "standardize the UI end to end", "onboard then clean the UI", "design prim", "/design-prim".
Removes AI tone from Chinese writing so text reads as human-written rather than model-generated. Covers blogs, technical documentation, tweets, release notes, scientific papers and other scenarios. Trigger words: 去 AI 味, 改写, 润色, 审稿, 翻译腔重, 像人写的, 自然一点. For removing AI tone in English, see deslop-en
Detect Pass-the-Hash (T1550.002) attacks by analyzing NTLM authentication patterns, flagging Type 3 logons using NTLM where Kerberos would be expected, and correlating with credential-dumping indicators. Use when threat hunting for lateral movement via stolen NTLM hashes, triaging EDR/SIEM alerts on suspicious NTLM logons, scoping compromise during incident response, or validating detection coverage in a purple team exercise.
(macOS, requires schedule-manager) Use this skill whenever the user wants to write a personal diary entry or daily journal — this includes any request to record today's events, write a diary, log what happened today, or capture personal reflections. Invoke immediately for phrases like 帮我写日记, 写日记, 记录今天, 今天的日记, "write my diary", or "daily log". This skill guides a complete journaling session: reviewing today's tasks from Reminders, reflective guided questions, composing the entry, and scheduling follow-up plans. Distinct from diary-note (quick append) and weekly-report (work summary).
Converts PRD and Tech Spec into a detailed, sequenced list of implementation tasks. Each task is a functional, incremental deliverable with its own test suite. Outputs tasks.md and individual task files. Use when the user asks to create tasks, break down work, or plan implementation from an existing PRD and Tech Spec. Do not use for PRD creation, tech spec creation, or actual code implementation.
Performs comprehensive code review by analyzing git diff, verifying conformance with project rules, validating test suites, and checking adherence to Tech Spec and Tasks. Generates a structured code review report with severity-classified findings. Use when the user asks for a code review, wants to validate code quality, or needs pre-merge verification. Do not use for QA testing, bug fixing, or task implementation.
Shows current progress of a PRD — completed tasks, next pending task, completion percentage, and missing artifacts. Use when the user asks for status, progress, what's next, or wants to resume work on a feature. Read-only skill — does not modify any files. Do not use for task implementation, review, or QA.
Comprehensive document creation, editing and analysis, with support for tracked changes, comments, format preservation and text extraction. Use when Claude needs to work with professional documents (.docx files): (1) create new documents, (2) modify or edit content, (3) handle tracked changes, (4) add comments, or any other document task
Builds and optimizes language model programs with DSPy (Stanford NLP), using Signatures, modules (Predict, ChainOfThought, ReAct, ProgramOfThought) and optimizers (BootstrapFewShot, MIPRO, BootstrapFinetune). Use when replacing hand-written prompts with declarative signatures, when automatically tuning prompts against training data and a metric, when building multi-stage RAG pipelines or ReAct agents, when building classifiers or structured-output programs, or when saving and evaluating optimized LM programs across providers such as Anthropic, OpenAI, or Ollama. Not for quick prototypes or simple chains; use manual prompting or LangChain instead.
Use dspy.RLM (Recursive Language Model) for reasoning over contexts too large to fit in an LLM's working window — entire codebases, long logs, massive documents, or multi-step data exploration that needs a sandboxed Python REPL. Use when the input is >100k tokens, needs recursive chunking, or benefits from the LLM writing and running code to probe data.
Deploys static HTML to a public URL instantly with no authentication required. Use when asked to "host this", "deploy this site", "get a public link", "share this HTML", "quick deploy", "publish this page", or any request to make an HTML file publicly accessible via URL. Supports self-contained HTML files with inline CSS/JS.
Effect v4 (beta) development and v3 → v4 migration guidance. Use when building new Effect v4 code or upgrading from v3, including API renames, behavior changes, before/after examples, and v4-specific patterns for services, layers, generators, Effectable, error handling, and schema codecs.
Create sophisticated, Apple-inspired web interfaces with refined aesthetics and corporate elegance. Emphasizes generous whitespace, ultra-precise typography, asymmetric layouts, and purposeful interactions. Combines minimalist design philosophy with premium brand execution. Use when building premium landing pages, marketing sections, product showcases, or corporate interfaces that demand visual sophistication beyond standard business UI.
Pre-send email quality review and scoring across 5 dimensions (subject line, copy quality, technical/HTML, deliverability, compliance). Analyzes subject lines, body copy, HTML structure, spam triggers, and CAN-SPAM compliance. Scores 0-100 with detailed recommendations for improvement. Use when user wants to review an email before sending, check email quality, or validate marketing email best practices.
Designs complete email automation sequences with timing, subject lines, copy, and conditional logic. Supports welcome series, nurture campaigns, re-engagement, abandoned cart, post-purchase, review requests, and custom sequences. Adapts sequence type, cadence, frameworks, and conditional branching to business context. Use when user wants to create an automated email series triggered by subscriber actions or time intervals.
Compose high-converting emails using proven copy frameworks (PAS, AIDA, BAB, FAB, 4Ps). Generates subject line variants with scores, responsive HTML templates with dark mode support, plain-text fallback, and preheader recommendations. Optimized for cold outreach, newsletters, product launches, promotions, and transactional emails. Adapts to user context from email-profile.md.
Guidance for Microsoft Entra Permissions Management (CIEM) — discovers, right-sizes, and monitors permissions across Microsoft Azure, AWS, and Google Cloud. Covers cloud onboarding, the Permission Creep Index (PCI), generating least-privilege policies from observed activity, on-demand permission grants, and workload identity coverage. WHEN: Entra Permissions Management, CIEM, multicloud permissions, Permission Creep Index, PCI, right-size permissions, unused permissions, least privilege across AWS GCP Azure, cloud entitlement management, over-privileged workload identity, IAM right-sizing, multicloud IAM posture. DO NOT USE for Azure-only RBAC role choice (use azure-role-selector) or Defender for Cloud security posture (use defender-for-cloud-hardening).
Use when a complex problem seems unsolvable and needs digging to the essence — diagnosis only, not fixing. Symptoms include massive logs with no root cause surfaced, complex crashes, architectural essence analysis, requirement authenticity verification, systemic chronic issues, information overload, disconnect between symptoms and essence, multiple intertwined factors, or premature-convergence tendencies — even when the user never explicitly says "diagnose". Also triggers on 本质诊断, 根因诊断, 深度诊断, 诊断问题本质, 梳理问题逻辑, 证据链分析, 逻辑链分析, essence-diagnosis.
Use for ethical thinking, or when a plan could harm or unfairly burden people: "should we do this?", "is this fair?", "privacy or bias risk", "who gets hurt?". Maps stakeholders, harms, power and consent. Skip for legal advice, neutral fact lookups and implementation-only work.
Analyzes Copilot Studio evaluation results using Practical Guidance on Agent Evaluation's 10-step playbook (Steps 6, 7, and 9) plus Microsoft's triage diagnostics. Returns a gate-based SHIP / ITERATE / BLOCK verdict with root cause classification, remediation, and pattern analysis.
Evaluates code generation models across HumanEval, MBPP, MultiPL-E, and 15+ benchmarks with pass@k metrics. Use when benchmarking code models, comparing coding abilities, testing multi-language support, or measuring code generation quality. Industry standard from BigCode Project used by HuggingFace leaderboards.
Build a reporting or executive presentation where every slide title states a conclusion and the deck follows a pyramid, answer-first structure. Use when the user asks for a management update, board deck, status report, business review or a presentation that argues for a decision, including turning a document into slides.
Explain content or any concept inside it at the depth the user needs — ELI5, practitioner level, or expert deep-dive — with a jargon glossary and context the original assumes. Use when the user says "explain this", "what does X mean here", "break this down", "I don't understand this part", or shares content and asks how something in it works.
Methodology for exploratory data analysis on scientific files. Decision frameworks by data type (tabular, sequence, image, spectral, structural, omics), quality assessment, report generation, format detection across 200+ formats. Use when given a data file for initial exploration or to pick an analysis before a pipeline.
Use this skill to move files between host and device with `adb pull` and `adb push`, including the modern `-z` (compression), `-Z` (no compression), `--sync`, and `-a` (preserve attrs) flags. Covers the path-permission rules — `/data/local/tmp/` is freely writable, `/sdcard/` (alias `/storage/emulated/0/`) is shell-writable with scoped-storage rules on API 30+, `/data/data/<pkg>/` requires `run-as <pkg>` on debuggable builds, and `/sdcard/Android/data/<pkg>/files/` is package-owned but pullable. Includes `adb shell run-as <pkg> cat <path>` for text grabs, `adb exec-out run-as <pkg> tar cf - <path> | tar xf -` for binary-clean directory grabs, the `connected_android_test_additional_output/` Gradle output dir, and the `androidx.test:services` `useTestStorageService` flag for the modern artefact-collection API. If the user mentions "adb pull permission denied data data", "run-as", "scoped storage shell", "tar through adb", "test storage service", or "exec-out vs shell for binary", use this skill.
Dual skill for deploying scientific models. FastAPI provides a high-performance, asynchronous web framework for building APIs with automatic documentation. Streamlit enables rapid creation of interactive data applications and dashboards directly from Python scripts. Load when working with web APIs, model serving, REST endpoints, interactive dashboards, data visualization UIs, scientific app deployment, async web frameworks, Pydantic validation, uvicorn, or building production-ready scientific tools.
Apply when deciding, designing, or implementing data fetching with FastStore GraphQL files in src/graphql/ or src/fragments/, or configuring faststore.config. Covers API extensions, GraphQL fragments, server-side and client-side data fetching, and custom resolver patterns. Use for integrating custom data sources or extending the FastStore GraphQL schema.
Build UI interactions whose motion comes from a physical model instead of durations and easing curves. Turn a feel described in words ("a toggle that feels heavy", "a drawer with a magnetic snap", "a card you throw into a corner", "a knob with heavy detents", "a sticky slider") into a production component driven by mass, stiffness, damping and force fields (magnets, friction, walls), on a line, in a plane or around a dial, plus a test rig that plots position, velocity and the force field, overlays the model's prediction on the measured motion, and exports the same spring to Framer Motion, react-spring, SwiftUI, Compose or CSS linear(). Use this whenever someone asks for an animation, transition, spring, gesture, drag, flick, snap points, bottom sheet, toggle, slider, carousel or any micro-interaction that should "feel" a certain way, or wants to tune or explain a spring, even if they never say "physics".
Complete Feishu (Lark) Skills collection for Claude Code / OpenClaw. Includes 10 skills covering: document management, messaging, spreadsheets, Bitable, interactive cards, bot bridge, cross-group memory, and leave requests. 飞书 Skills 全集，含文档管理、消息发送、表格操作、多维表格、卡片消息、 Bot 桥接、跨群记忆、请假审批等 10 个 Skill，附完整配置教程。 Trigger on: feishu, 飞书, lark, feishu doc, feishu message, feishu sheet, feishu bitable, feishu card, feishu bridge, feishu bot
Fetch and normalize any content source into clean text with metadata — YouTube video transcripts, TikTok captions, web articles, PDFs, tweets/X posts, local files. Use when the user shares a YouTube link, TikTok link, article URL, tweet/X link, or PDF (URL or file) and you need its actual text content to summarize, analyze, fact-check, or answer questions about it.
Authoritative Python-FFmpeg parameter integration reference ensuring type safety, accurate parameter mappings, and proper unit conversions. PROACTIVELY activate for: (1) ffmpeg-python library usage, (2) Python subprocess FFmpeg calls, (3) Caption/subtitle parameter mapping (drawtext, ASS), (4) Color format conversions (BGR, RGB, ABGR, ASS &HAABBGGRR), (5) Time unit conversions (seconds, centiseconds, milliseconds), (6) Type safety validation (int, float, string), (7) Coordinate systems, (8) Parameter range enforcement, (9) Frame pipe handling, (10) Error detection for type mismatches. Provides: Complete parameter type reference, color format conversion tables, time unit conversion formulas, validation patterns, working Python examples with proper typing.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.