Skills
101,323 packaged capabilities and playbooks for agents. Install any of them with one command.
All
Manages permanent memory storage for decisions, blockers, context, preferences, and procedures. Use when user says "remember", "save this decision", "what did we decide", "recall", "search memories", "any blockers", or when making important architectural decisions. Provides SDAM compensation through external memory.
For CRD proposers working across projects: draft or revise requirements, consolidate recipient responses, and continue Q&A. Recipients should use crd-respond for assessment and replies. Not for Kubernetes CRDs, code review, or direct implementation. Clarify your side's position and converge on a contract while preserving the rationale from each exchange.
Creates comprehensive GitHub pull requests with automated quality validation and security scanning, then hands off to /github:review-pr for CI monitoring and reviewer-comment triage. This skill should be used when the user asks to "create a PR", "submit a pull request", or needs to merge completed work with full compliance checks.
Validates and repairs an in-scope change, publishes its draft pull request, reconciles metadata and CI, and marks it ready for review. Use when asked to open, raise, or publish a pull request for the current change, or when the user runs /create-pr.
Author, review, or refactor a skill for the SAP Fiori Tools / open-ux-tools ecosystem. Use this skill whenever the user wants to create a new skill, rewrite an existing one, add references, split an oversized SKILL.md, enforce the `sap-fiori-<feature>` naming convention, validate frontmatter, add CAP/RAP samples, declare OData V2 vs V4 requirements, mark a skill as internal (developer-facing) or external (public), or produce a verification checklist — even if the user does not explicitly say "skill". Enforces host-agnostic tooling, canonical fallbacks, explicit scope boundaries, and progressive-disclosure chunking (<500 lines per file).
Build a callout-annotated CRM or sales-tool training video from a screen recording of a rep or enablement lead using Salesforce, HubSpot, or an internal deal-desk tool. Redacts real customer and deal data first, identifies the exact fields and buttons that are the real teaching points on a dense CRM screen, adds arrows, callouts, highlights, and zoom-ins pointing precisely at them, and pairs each with narration explaining not just what to click but why. Use when the user says "turn this CRM screen recording into a training video", "make a callout-annotated walkthrough of our Salesforce process", or "build a tool training video from this screen capture".
Use when an approved campaign, launch, research finding, or asset needs a deliberate plan across several social channels. Trigger phrases include "adapt this for LinkedIn and X", "where should this campaign go?", "make a distribution plan", "repurpose this without duplicating it", and "what is each platform's role?". Do not use this to invent a campaign strategy, bulk-post identical copy, or schedule a calendar; use campaign-brief for the strategic source, content-repurposer for asset transformations, and content-calendar for dates.
Use when the user wants to drive the CrowdSec Console **Service API (SAPI)** — the premium cloud REST API at admin.api.crowdsec.net — to programmatically manage blocklists (add/remove/bulk IPs, share, subscribe engines), allowlists, firewall/appliance integrations (Palo Alto, Fortinet, Cisco, F5, Sophos, pfSense/OPNsense…), remediation ROI metrics, and org-level decisions. Acts on the user's behalf with their API key. This is the cloud/API skill — for the local engine, cscli, and bouncers use the `crowdsec` skill.
CodeStable: understand and drive software evolution with Vision / Project Spec / Epic / Issue and codestable/ institutional memory; act according to the user's current stance—discussion, quick fixes, managed implementation, bug fixing, organizing vision/spec, understanding the status quo, closing out, or review as needed. Triggers: cs, CodeStable, discussion/talk it through first, quick/quick fix/dive right in, spike, fix bug/debug, organize vision/spec, close/wrap up, review, how does this system work, or the project already has codestable/ and is working on vision, spec, bug, implementation, and closing.
NCU-driven iterative optimization workflow for CUDA/CUTLASS/Triton/CuTe DSL kernels. MANDATORY: every optimization MUST start with NCU profiling, followed by multi-dimensional analysis, then targeted code modification, then re-profiling to verify. Supports roofline, memory hierarchy, warp stalls, instruction mix, occupancy, divergence analysis. Provides implementation-specific code modifications: Native CUDA (launch config, memory patterns, async copy, Tensor Core), CUTLASS (ThreadblockShape, stages, epilogue, schedule policy, alignment), Triton (autotune params, compiler hints, tl.* API patterns), CuTe DSL (threads_per_cta, elems_per_thread, tiled_copy, copy atom, shared memory, warp/cta reduce). Use when optimizing any CUDA kernel performance.
Deploy an existing project to a website on Hostinger web hosting (Shared, Cloud, or Agency plans) and keep it deployed: choose the right deployment for static sites, Node.js apps (Next.js, Nuxt, Express, Vite, and similar), PHP apps, and WordPress plugins or themes; set up auto-deploy from GitHub or GitLab; manage Node.js environment variables and MySQL databases; and verify the live site. Triggers: deploy this project, deploy to Hostinger, push my app live, redeploy, set up auto-deploy, connect my GitHub repo, add environment variables, my app needs a database, deploy my WordPress plugin or theme
Reverse-engineer a website's internal API by recording browser traffic into a HAR file, then generate a standalone client or CLI that calls the endpoints directly, with no browser needed after the first recording. Use when asked to "derive a client", "build a CLI for <site>", "reverse engineer this site's API", "record network requests", "turn this site into an API", or when the same site will be automated repeatedly and direct HTTP calls would beat driving the browser every time.
Greenfield design pipeline for any digital product (landing pages, websites, web apps, mobile apps): extract content, author a single-file visual identity in .agents/design/ (DESIGN.md, tokens plus rationale), define layout or screen flow, preview and refine designs, push to an external design tool when available. Use when building pages or app screens from references, images, briefs, or an existing codebase; defining layout or screen flow; previewing and tuning designs; authoring or refreshing DESIGN.md from a source. Triggers: "extract copy", "extract design", "extract from codebase", "web capture", "wireframe", "screen flow", "layout the page", "preview the design", "creative variants", "tune design", "mockup", "redesign this app", "modernize this app", "brand refresh", "validate DESIGN.md", "refresh design tokens", "push design to external tool", "generate hero banner", "create product image". Not for feature implementation spec, system architecture, or PR/code review.
Analyzes your recent Claude Code chat history to identify coding patterns, development gaps, and areas for improvement, curates relevant learning resources from HackerNews, and automatically sends a personalized growth report to your Slack DMs
Build and maintain Dexie.js data layers for IndexedDB-backed offline-first web applications. Use when tasks involve Dexie setup, schema/index design, CRUD/query implementation, transactions, live queries, migrations, or debugging in Vanilla JS, React, Next.js, TanStack Start, Svelte, or SvelteKit.
Choose the best delivery mechanism for an offer — Done For You, Done With You, or Do It Yourself. Use when the offer exists but feels weak, delivery format is unclear, or scaling vs customization is a tradeoff. Scores each model on speed, ease, perceived value, and scalability.
Designs structured interview guides, survey instruments, and JTBD probes to learn from users while avoiding common research biases (leading questions, confirmation bias, selection bias). Use when validating product assumptions before building, discovering unmet user needs, understanding customer problems and workflows, testing concepts or positioning, researching target markets, identifying jobs-to-be-done and hiring triggers, or uncovering pain points and workarounds.
Extract plain text from a document file - PDF, DOCX, XLSX, PPTX, RTF, or plain text/markdown/HTML. Use when a binary document needs to be turned into text, for example a contract PDF or an EHR DocumentReference attachment. Other skills (fhir) invoke scripts/extract.ts directly; the contracts MCP server bundles its own copy (servers/documents/src/extract.mjs) so its bundle stays self-contained — port fixes to both.
Guidance for Docker image registries — publishing images to Docker Hub, pulling images, tag management strategies, private registry setup (Docker Registry container and Harbor), registry mirrors for acceleration, image signing and content trust, and registry API usage. Use when the user asks about docker hub, image registry, docker push/pull, private registry, docker login, tag management, Harbor, or needs to manage container images in registries. 使用场景：docker hub、镜像仓库、docker push、docker pull、私有仓库、registry、Harbor、镜像标签管理.
Guidance for running and managing Docker containers. Covers the complete container lifecycle (create/start/stop/rm/restart), resource constraints (--memory/--cpus), health checks (HEALTHCHECK), restart policies, logging drivers, container inspection (inspect/stats/top), exec debugging, and signal handling. Use when the user asks about docker run, container management, resource limits, health checks, docker logs, docker exec, restart policies, or needs to manage running containers. 使用场景：docker run、容器管理、资源限制、健康检查、重启策略、docker logs、docker exec、docker ps.
Real-time Douyin hot topics across the platform, for content creators, operations, e-commerce and marketing teams when users want to know what is trending on Douyin right now; generates real-time hot topics, current hot list content and topics worth following up on immediately from the input material.
Drafts or rewrites a Related Work / prior work section positioned against actually-retrieved papers — derives the REQUIRED clusters from the paper's own claimed scope (not just whatever was retrieved), clusters prior work into themes, articulates the delta (what this paper adds) per cluster, enforces a per-cluster citation floor and routes empty expected clusters back to find-papers, follows the target community's placement and citation conventions (numeric ACM/IEEE vs natbib author-year, single- vs double-blind self-citation, where the section sits at NeurIPS/CHI/SIGMOD-style venues), and admits only verified citations. Use when the user says "related work", "prior work", "position my paper against the literature", "how do we differ from X", "reviewers said related work is thin/missing/a laundry list", or asks to add citations to a draft. Works from papers found via find-papers and/or an existing .bib — it never invents references.
Provides comprehensive guidance for draw.io flowcharts including flowchart creation, shapes, connectors, and diagramming. Use when the user asks about draw.io flowcharts, needs to create flowcharts, design process diagrams, or visualize workflows.
Portable Effect patterns for robust promise execution. Use when wrapping async operations with timeouts, retries, tagged errors, caching, concurrency, pattern matching, or tracing - all designed to resolve to a plain Promise via Effect.runPromise.
Reduce friction and effort in an offer to increase its perceived value. Use when users feel overwhelmed, drop-off is high, or the offer feels too complex. Maps friction points, removes unnecessary steps, adds templates, and suggests automation and done-for-you elements.
Owns data understanding before any model is designed. Place `data_analysis/data_analysis.py` and do not execute it. Run `materialize.py` once, write `data_analysis.md` and JOURNAL § Data understanding. Never design the model, edit `src/<pkg>/`, or modify raw data files. TRIGGER when the user asks to explore, profile, or understand the data; triage sent them here (`status.data_analysis` missing); a data source changed; they want to refresh a recorded EDA; or leakage or research arises on a recorded EDA. STOP when `status.setup.pending` is non-empty (load `setup-ml-project`), when there is no data (send to triage), the request is not raw-data exploration, or EDA is recorded with no refresh and no methodology concern. Do not invent a package root or a root `JOURNAL.md` while setup is pending. HOW TO USE: G-TABULAR via `add-python-package`, infer or ask the targets, load `plot-ml-figure` if installed, then ask the five-option continuation board, including Close.
Complete guide to FFmpeg filter_complex syntax, multi-input/output pipelines, and composition patterns. PROACTIVELY activate for: (1) filter_complex syntax questions, (2) Multi-input video composition (PiP, grids, overlays), (3) Video transitions (xfade, fades, wipes), (4) Audio mixing (amix, amerge), (5) Stream splitting and routing, (6) Named stream labels, (7) Complex filtergraphs, (8) GPU + CPU hybrid filter chains, (9) ABR encoding ladders, (10) Text and watermark overlays. Provides: Syntax reference, composition patterns, audio mixing examples, transition effects, GPU-accelerated filtergraphs, debugging tips.
Provides comprehensive guidance for Flask framework including routing, templates, forms, database integration, extensions, and deployment. Use when the user asks about Flask, needs to create Flask applications, implement web routes, or build Python web applications.
Use when debugging or preventing errors in Frappe Server Scripts. Prevents ImportError (the #1 error), NameError for restricted builtins, sandbox violations, doc_events not firing, wrong script type selection, SQL injection, permission denied in scheduled scripts, infinite loops, and API scripts not returning JSON. Covers error message mapping table. Keywords: server script error, ImportError, NameError, sandbox, restricted, ImportError in server script, script not running, sandbox error, restricted function, frappe.throw, doc_events, scheduler, API script, SQL injection.
Use when implementing server-side features via Setup > Server Script: document validation, auto-fill, API endpoints, scheduled tasks, permission queries. Covers sandbox-safe coding, script type selection, testing, migration to controllers. Keywords: how to implement server script, which script type, sandbox limitation, Document Event, API script, Scheduler Event, Permission Query, migrate to controller, no-code automation, run code on save, auto-fill field, server-side validation, scheduled script.
Use when creating or customizing Workspace pages in Frappe v14-v16. Covers Workspace DocType structure, shortcuts, number cards, dashboard charts, custom HTML blocks, JSON content format, shipping workspaces with custom apps, and role-based access control. Prevents common mistakes with content/child-table desync and missing fixtures. Keywords: workspace, desk, dashboard, number card, chart, shortcut, customize desk, dashboard setup, add shortcut, module page, sidebar customize, workspace builder, module, fixtures, sidebar.
Use when working with Frappe Cloud, Press API, provisioning sites, or managing benches on Frappe Cloud infrastructure. Prevents failed deployments from misconfigured cloud settings and API misuse. Covers Frappe Cloud dashboard, Press API, site provisioning, bench management, environment variables, cloud-specific limitations. Keywords: Frappe Cloud, Press, cloud API, site provisioning, bench management, FC, frappecloud, Frappe Cloud deploy, FC hosting, cloud setup, managed hosting.
Use when configuring scheduler events and background jobs in Frappe/ERPNext v14/v15/v16. Covers scheduler_events in hooks.py, frappe.enqueue() for async jobs, queue configuration, job deduplication, error handling, and monitoring. Keywords: scheduler, background job, cron, RQ worker, job queue, async task, frappe.enqueue, scheduled task, cron syntax, how often does it run, background job example, enqueue example.
Reclaim disk space on macOS by auditing and clearing build artifacts, package-manager caches, stale git worktrees, and offloading large media to an external drive. Use whenever the user complains about low storage, a full disk, "out of space" errors, or asks to clean up, offload, or free up their machine — even if they don't name specific folders.
Guides an AI to implement a single-symbol base-price grid program on Futu OpenAPI: validate six parameters first, then write a restart-safe limit-order reconciliation loop and accept it on a paper account. Use when the user wants a grid strategy, grid program, quantitative grid, or grid trading bot, or says 网格策略、网格程序、量化网格、网格交易机器人、实现网格 or 帮我写一个网格. Request fields follow the OpenAPI llms.md. This skill does not repeat request bodies. Paper account by default.
Use this skill whenever the user’s main ask is news, events, listings, social/UGC, or market move attribution. Primary gate-cli news (optional info for market context). Covers briefings, event explain, market move attribution via explain-market-move, exchange announcements, UGC X/Reddit/YouTube, sentiment. Triggers: what happened, why crash, why pump, market move reason, new listings, community take. Delegate: gate-info-research (research-first), gate-info-web3 (on-chain), gate-info-risk (safety). v0.7.6+ news shortcuts (+brief, +event-explain, +community-scan); legacy fallback below 0.7.6.
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.