Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 79 870 files from 1 769 authors, of which 62 217 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
Generate a custom checklist for the current feature based on user requirements.
Execute the implementation planning workflow using the plan template to generate design artifacts.
Create or update the feature specification from a natural language feature description.
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
Create or update the project constitution from interactive or provided principle inputs, ensuring all dependent templates stay in sync.
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
Generate an actionable, dependency-ordered tasks.md for the feature based on available design artifacts.
Execute the implementation plan by processing and executing all tasks defined in tasks.md
Convert existing tasks into actionable, dependency-ordered GitHub issues for the feature based on available design artifacts.
Use for the subagent-as-a-tool pattern.
Use when connecting AI SDK for Python streams to AI SDK UI useChat clients.
Use when building serverless AI SDK for Python endpoints, handling hook approvals, deferring hooks, or resuming runs across requests.
Use when building custom agent loops. Modify tool dispatch, history management, hooks, control flow.
Use when adding durable execution to AI SDK for Python, building durable agent loops, or serializing messages across workflow steps.
Use for implementing custom providers in AI SDK for Python.
Use for AI SDK for Python async-generator tools, streaming tool output, subagent tools, PartialToolCallResult events, and custom tool aggregation.
Use for AI SDK for Python basics. Configure a model, make messages, stream, declare tools, build a basic agent.
Use when drafting or editing a post (toot) for Mastodon / the fediverse.
Use when drafting or editing a post or thread for X (Twitter): voice, hooks, length, threads, hashtags, and links.
Use when drafting or editing a post for Bluesky.
Use when drafting or editing a LinkedIn post in the house voice.
Use when drafting or editing a post for Threads (Meta): casual, conversational, reply-driven voice.
Generic writing-quality rules for any prose meant for humans to read: keep it natural, plain, and free of AI-sounding phrasing, and check it against the AI-phrases and plain-English reference lists. Load before drafting, editing, or reviewing content. Not needed for code or configuration work.
How to compose or revise the weekly digest email of a repository's open issues: subject line, grouping, one-line issue summaries, needs-attention and stale criteria, citation format, and the closing invitation to reply. Load when composing this week's digest or reworking a draft of it. Not for other emails, replies, or GitHub and Linear comments.
Playbook for triaging GitHub issues: checking for duplicates, applying the repo's existing labels, deciding whether to ask for a reproduction or close, and choosing when to act directly versus report back first. Load whenever asked to triage, label, dedupe, or close issues, whether the request came from a GitHub mention, an email reply, or a Linear delegation. Not needed for composing the weekly digest, answering general questions on a thread, or creating Linear issues.
Conventions for bridging GitHub and Linear: creating Linear issues from GitHub issues, cross-referencing the two trackers, and checking whether a GitHub issue is already tracked in Linear. Load when asked to create Linear issues from GitHub issues, link the trackers in either direction, or report tracking status. Not needed for work that stays inside one tracker.
Cross-resource deep analysis of a specific distributed trace. Takes an operation ID and correlates telemetry across multiple Application Insights resources to build a unified picture of a distributed operation.
Guide users through enabling the Application Insights Profiler for .NET on their platform. Use this when asked to enable the profiler, set up profiling, or when profiler data is missing.
Download a profiler trace file from Application Insights Profiler. Use this when asked to download, save, or export a profiler trace or .etl/.nettrace file.
Download a snapshot dump file from Application Insights Snapshot Debugger. Use this when asked to download, save, or export a snapshot or .dmp file for offline debugging.
Fetch and display the hot path from an Application Insights Profiler trace. Use this when asked to get the profiler hot path, call tree, or trace analysis for an App Insights resource.
Analyze AI agent telemetry from Application Insights, including anomaly detection, trend analysis, and performance statistics. Use this when asked to analyze AI agent performance, detect anomalies in agent telemetry, or review agent trace data.
Explore errors in Application Insights — exceptions, failed requests, and failed dependencies. Use this when asked to find errors, investigate failures, identify reliability issues, or recommend which errors to fix.
Guide users through enabling the Application Insights Snapshot Debugger for .NET on their platform. Use this when asked to enable the snapshot debugger, set up snapshot collection, or when snapshot data is missing.
Fetch exception details, call stack, and variable values from an Application Insights Snapshot Debugger snapshot. Use this when asked to inspect a snapshot, view exception info, debug an exception, or see variable values from a crash.
Guide for analyzing performance issues based on profiler and code optimizations, including CPU, latency, and throughput. Use this when asked to investigate performance bottlenecks or optimize application performance.
Adds a new extension point route to a Sitecore Marketplace app. Use when the user wants to add a custom field, dashboard widget, context panel, fullscreen page, or standalone route.
Scaffolds a new Sitecore Marketplace app using the official SDK. Use when the user wants to create a new marketplace app, start a new Sitecore project, or initialize a marketplace integration.
Builds UI components using the Blok design system for Sitecore Marketplace apps. Use when the user wants to create UI, add components, build a page layout, or work with Blok/shadcn components in a marketplace app.
Sitecore Marketplace SDK API reference. Use when the user asks about SDK methods, types, queries, mutations, subscriptions, or how to use any Sitecore Marketplace SDK API.
Adds XM Cloud API integration to a Sitecore Marketplace app. Use when the user wants to access Sites, Pages, Authoring, Content Transfer, Search, or Agent APIs from XM Cloud.
Deploys a Sitecore Marketplace app to Vercel with correct CSP headers and configuration.
Adds AI Skills integration (Brand Review API) to a Sitecore Marketplace app. Use when the user wants to add AI-powered content analysis, brand review, or brand compliance checking.
Deploy applications and websites to Vercel with a claimable preview. Use when the user requests deployment actions like "deploy my app", "deploy and give me the link", "push this live", or "create a preview deployment". No authentication required - returns preview URL and claimable deployment link.
Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria. Based on https://github.com/multica-ai/andrej-karpathy-skills.
Best practices for Remotion - Video creation in React
Project knowledge for making this repo preview correctly from Vercel Devbox/Sandbox. Use when configuring Next.js dev settings, explaining how users preview the app from Devbox, or responding to localhost, Sandbox URL, port 3000, or cross-origin dev-resource issues.
Python `ai` SDK — models, providers, streams, events, tools, agents, hooks, MCP, AI SDK UI, structured output, and media generation
Project knowledge for configuring Vercel Services in this Next.js and FastAPI starter. Use when editing vercel.json, adding services, debugging frontend-to-backend routing, or explaining Vercel Services local development and deployment.
AWS Aurora PostgreSQL and Aurora DSQL Python integration using asyncpg. Use when building Python backends with Aurora PostgreSQL or Aurora DSQL.
Use when building Hono web applications deployed to Vercel, or when the user asks about Hono APIs, routing, middleware, JSX, validation, testing, or streaming. TRIGGER when code imports from 'hono' or 'hono/*', or user mentions Hono. Use `npx hono request` to test endpoints.
Fetches LangSmith traces for debugging agent behavior. Use when troubleshooting agent issues, reviewing conversation history, or investigating tool calls.
Creates code-based evaluators for LangSmith-traced agents. Use when building custom evaluation logic, testing tool usage patterns, or scoring agent outputs programmatically. Triggers on requests to evaluate agents, create evaluators, or run experiments against LangSmith datasets.
Get an unseen (possibly old or abandoned) TypeScript/JavaScript GitHub repo building in a disposable sandbox. Detects the era-appropriate toolchain, runs the documented setup, then drives Setup/Compile/Test/Runtime to green with a progress-based convergence loop. Use when you need to bring up a repo you've never seen and honestly report what it took. Scope is TS/JS (Node + package manager + tsc) only.
How to make minimal repos that the TypeScript team enjoys working with
Draw 2D shapes (filled rectangles, outlined frames, ASCII bitmaps) on the 32×32 MCP pixel grid. Use this skill whenever the user asks to draw a square, rectangle, box, frame, outline, border, sprite, or other 2D shape on the pixel grid, or when the task involves placing a block or bitmap at specific grid coordinates. The pixel-grid MCP server only exposes a 1D `write(offset, bits)` primitive; this skill compiles a 2D drawing into the sequence of `write` calls that implements it.
Writing-quality guardrails for any prose the agent drafts or edits: blog posts, page copy, Notion drafts, release notes, marketing text. Use this skill whenever writing or revising content meant for humans to read, to keep the prose natural, plain, and free of AI-sounding phrasing. Not needed for code, queries, or structural CMS work.
HUMAN-ONLY shepherd design approval. Run after reviewing .shepherd/2-design.md, .shepherd/3-success-criteria.md, and .shepherd/_panel.json. Records the approved panel in _state.json, writes .shepherd/_design.approved, and hands control back to /shepherd. The agent cannot invoke this.
HUMAN-ONLY shepherd create-PR approval. Run after reviewing the change and .shepherd evidence. Writes .shepherd/_create_pr.approved and hands control back to /shepherd. The agent cannot invoke this.
Post-run analysis of one or more shepherd runs in the current session — per-stage cost, signal-per-stage counterfactuals, reviewer overlap, gate calibration, and ranked improvements. Run after a run reaches phase=done. Invoke as /shepherd-evaluate.
Claude Skills — questions
Answers built from the skills we actually parsed.