Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Development
Technical documentation discovery via context7 and web search. Capabilities: library/framework docs lookup, topic-specific search. Keywords: llms.txt, context7, documentation, library docs, API docs. Use when: searching library documentation, finding framework guides, looking up API references.
Professional DrissionPage (Python) automation development workflow and troubleshooting using the bundled DrissionPage docs (v4.1.1.2); use when writing/debugging/refactoring DrissionPage scripts for ChromiumPage/WebPage/SessionPage, element locating and interaction, waits, downloads/uploads, network listening, and configuration.
Drizzle ORM v1 RC upgrade and migration patterns. Use when upgrading drizzle-orm/drizzle-kit to v1 (beta), migrating from Relational Queries v1 to v2, updating relations schema definitions, or working with new v1 features like `through` for many-to-many relations and object-based `where`/`orderBy` syntax.
Design, refactor, and review Effector state management using modern v23+ patterns. Use when tasks involve createStore/createEvent/createEffect modeling, dataflow with sample/attach/split, scope-safe SSR with fork/allSettled/serialize/hydrate, React integration with useUnit, Solid/Vue integration patterns, fixing scope loss, or replacing anti-patterns such as business logic in watch, imperative calls in effects, and direct getState business reads.
Guide to building conversational AI voice agents on the ElevenLabs Agents Platform (ASR, LLM, TTS, turn-taking model). Covers agent configuration and system prompts, workflows, multi-voice, pronunciation dictionaries and speed control, knowledge base RAG, client/server/MCP/system tools, Scribe real-time STT, WebRTC and WebSocket connections, testing and evaluation, analytics, cost optimization, GDPR/HIPAA/SOC 2 compliance, and agents-as-code CLI workflows. Includes production templates for @elevenlabs/react, @elevenlabs/client, @elevenlabs/react-native, Next.js, Swift and Cloudflare Workers, plus 17+ documented fixes: deprecated @11labs packages, Android audio cutoff, CSP violations, missing dynamic variables, case-sensitive tool names, webhook authentication failures and WebRTC misconfiguration. Use when building, configuring or debugging an ElevenLabs voice agent.
Enforce portable Elysia app architecture: feature modules, folder trees, naming, one-route-one-file, thin handlers, schema/services ownership, mount tables, and direct imports of db/runtime (no route factories). Use when scaffolding a new Elysia API or feature, adding endpoints, reviewing or reorganizing module layout, or when the user asks for elysia-architecture / Elysia house style. Reject export function routes(db) / decorate-db injection. Optional with-* overlays for Effect, env, session auth, capability authz, API keys, Drizzle, OpenAPI, observability, HTTP errors, domain packages, plugins, cron, webhooks, service auth, and Vitest. HTTP tests: Eden treaty(app), not handle/Request.
Perform comprehensive, deep analysis of a system and its subsystems to identify bugs, race conditions, stale documentation, dead code, and correctness issues. Use when asked to "audit this system", "exhaustive analysis of X", "analyze for correctness", "root out issues in...", "deep dive into...", "verify this code is correct", "find bugs in...", or when reviewing agent-written code for production readiness. Automatically decomposes systems into subsystems, applies appropriate analysis checklists, and produces structured findings with severity classification.
Use when the user wants an autonomous ML research loop that explores the space broadly rather than hill-climbing one approach. A temperature scheduler replaces the usual hypothesis step: it forces several wild, diverse swings (full rewrites, different architectures/regimes) early, then enters an adaptive phase that picks swing / merge / exploit per iteration — with a hard stagnation guard that bans further small-step exploits once they run too long, forcing a pivot back to a swing or merge. Tracks an approaches.md registry and a move_type per iteration; analyses every run before the next move. One change per iteration; loops forever until interrupted. Not for the standard analysis-first ml-autoresearch (which lets analysis alone choose each change), one-off training runs, or sweeps.
Runs Fastly Compute WASM binaries locally and serves as the authoritative reference for Compute platform internals. The fastlike source code is highly readable and covers the host ABI, caching and purging APIs, KV/config/secret store interfaces, rate limiting with counters and penalty boxes, ACL lookups, the full request lifecycle, backend fetch semantics, and a built-in per-request profiler with hostcall spans, backend waterfalls, native CPU samples, and optional deep metrics (body bytes, cache outcomes, header summaries, wasm heap curve). Use when working with Compute runtime internals or host calls, understanding how edge data stores behave at runtime, profiling local Compute apps, or testing WASM binaries locally. Prefer this skill over Viceroy for any non-Rust Compute work — its source code is easier to understand as a Fastly Compute API reference.
Builds MCP (Model Context Protocol) servers in Python with FastMCP, covering tool, resource, and prompt definitions, storage backends (memory, disk, Redis, DynamoDB), server lifespans, the 8 built-in middleware types, server composition via import and mount, OAuth Proxy and authentication, OpenAPI integration, client configuration, and FastMCP Cloud deployment. Includes templates for basic servers, storage, middleware, composition, OAuth proxy, API integrations, testing, and production architectures, and prevents 25+ common errors such as storage misconfiguration, lifespan and middleware-order mistakes, circular imports, module-level server issues, async/await confusion, OAuth security holes, and cloud deployment failures. Use when creating or debugging a Python FastMCP server, defining mcp tools, resources, or prompts, wiring fastmcp storage or middleware, mounting sub-servers, setting up MCP OAuth, or deploying to FastMCP Cloud.
Analyze and optimize FiveM and RedM resources for client/server resmon, Lua/native CPU, Wait(0) loops, hitch warnings, TriggerServerEvent / TriggerClientEvent / TriggerLatent* spam, payload size, broadcasts (-1), state bags, first-GET cache, MySQL load-at-start, neteventlog, profiler, and bandwidth. Use when reviewing, refactoring, debugging, or optimizing FiveM/RedM Lua performance or network traffic. Use when the user runs /fivem-redm-performance-network or $fivem-redm-performance-network.
Audit and fix HTML metadata including page titles, meta descriptions, canonical URLs, Open Graph tags, Twitter cards, favicons, JSON-LD structured data, and robots directives. Use when adding SEO metadata, fixing social share previews, reviewing Open Graph tags, setting up canonical URLs, or shipping new pages that need correct meta tags.
Archive or delete a stale GrowthBook feature flag, walking the user through inlining the flag's effective value at call sites in the codebase before removal. Use when the user says "delete this flag", "remove this stale flag", "clean up flag X", "archive this flag", "we don't need this flag anymore", or "get rid of this flag and its experiment-ref rule". For finding stale flags first, use flag-search. For editing rules instead of removing the flag, use flag-targeting. For stopping an experiment that uses the flag, use experiment-stop.
Set up a monitored progressive rollout ("safe rollout") for a GrowthBook feature flag — combining a ramp schedule with guardrail metric monitoring, automated signals, and optional auto-rollback. Use when the user says "roll this out safely", "monitor the rollout with guardrail metrics", "set up a safe rollout", "I want to ramp this with automatic rollback if metrics regress", "configure monitoring on the ramp", "check the monitoring status of this rollout", "approve the next monitored step", or "roll back because guardrails are failing". For unmonitored ramps (just progressive coverage, no metrics), use flag-ramp directly. For simple on/off time windows, use flag-schedule.
Add, remove, or inspect feature-level prerequisites on a GrowthBook feature flag. Use when the user says "gate flag X on flag Y being enabled", "add a prerequisite", "flag X should only evaluate if flag Y is on", "remove the prerequisite on flag X", "what does this flag depend on", or "this flag should require the new-checkout flag to be true first". Feature-level prerequisites gate the entire flag — when the prerequisite flag is off, this flag returns its default value for every user. The prerequisite must be a boolean flag. For prerequisites scoped to a single rule (not the whole flag), use flag-targeting. For tracing the full dependency graph, use flag-graph.
Review React, Next.js App Router, TypeScript, Tailwind CSS, shadcn/ui, TanStack Query, and frontend UI code. Use when reviewing a PR, branch, diff, page, component, hook, Server Action, route handler, or API client; debugging frontend behavior; assessing production readiness; or checking runtime bugs, client/server boundaries, data fetching, security, accessibility, UX states, responsive layout, performance, SEO metadata, i18n, and maintainability.
Senior full-stack engineer mode. The agent owns technical decisions and implementation; the user is the engineering manager who sets the what/why and approves the Stage 3 plan before any code is written. Full-stack feature work in existing codebases: backend APIs, services, data models, migrations, background jobs, external integrations, CLIs, libraries, plus product UI (pages, forms, tables, modals, dashboards). Use when the user wants to implement / add / build / extend a feature: "let's build X", "add feature Y", "implement Z", multi-module refactors, schema changes, new endpoints, new pages or components in an existing product, or any non-trivial engineering task that requires reading existing patterns before writing code. Prefer over a plain coding response whenever the feature has more than one moving part.
Git + worktree orchestrator: status survey, per-worktree triage, commits, PRs, branches, releases, rebases — reads inline, writes to a background agent. Triggers on: git status, anything to commit, anything to push, commit, push, create PR, rebase, release, tag, changelog, worktree, land worktrees.
Analyze one GitHub user's code contribution activity in a specified UTC time window and generate a Markdown report focused on practical contributions (authored merged PRs, PRs merged by the target user, and authored commits) across accessible public and private repositories. Use when users ask to evaluate real engineering contribution over a period and want evidence links.
Implementation guide for Global Privacy Control (GPC) automated opt-out signal per CPRA Section 1798.135(e). Covers Sec-GPC HTTP header detection, JavaScript navigator.globalPrivacyControl API, and state-specific requirements for CA, CO, CT, MT, TX, and OR. Includes server-side detection code and compliance mapping.
Route goalflow (wanmol/goal-flow) work — a LangGraph framework that combines workflow graphs with agent loops — into exactly one mode: fit check, transpiling a Dify DSL export into runnable LangGraph Python, authoring workflow nodes and edges, building an `agent_kit` loop with middleware and a harness, wiring the serving layer (data adapters, SSE streaming, HITL, Redis/MySQL, API-key registration), or running the pre-publish security gate. Use when the user wants Dify's visual design without Dify's runtime, a graph node that hosts an agent loop, an OpenAI-compatible wire protocol over their own workflows, or prompt-injected `SKILL.md` capabilities. Triggers on: goalflow, goal-flow, dify to langgraph, dify transpiler, dify DSL export, BaseWorkflow, agent_kit, AgentBaseNode, DataAdapter, chunk processor, HITL interrupt, dify2langgraph. Route plain graph-API questions to `langgraph-fundamentals`.
Structure Go application logic with pragmatic CQRS. Use when command and query concerns are mixed, handlers are hard to test, one application service is becoming wide, or a Go service needs business-oriented command and query naming, separate handlers, thin ports, narrow consumer-owned interfaces, and guidance on when CQRS is or is not worth the cost.
Guide to the Google Gemini API using the current @google/genai SDK v1.27+, not the deprecated @google/generative-ai (sunset November 2025). Covers text generation, streaming, multimodal input (images, video, audio, PDFs), function calling including parallel and compositional calls, system instructions, multi-turn chat, thinking mode, generation parameters, context caching, built-in code execution, grounding with Google Search, error handling and rate limits, plus accurate 2025 model facts (Gemini 2.5 Pro, Flash and Flash-Lite with 1M input tokens, not 2M). Use when integrating Gemini, building multimodal or chat applications, deploying Gemini calls to Cloudflare Workers, migrating off @google/generative-ai, or debugging model not found, context window, function calling, or multimodal format errors. Embeddings live in the google-gemini-embeddings skill.
Analyze code dependencies, call graphs, and architecture using the graph-it CLI. Use this skill whenever the user wants to understand how their code is connected — even if they don't say "dependency graph" explicitly. Trigger for: "what calls this function", "who uses this class", "is it safe to delete X", "what breaks if I change Y", "show me the architecture", "trace the execution from main", "find circular imports", "give me an overview of this module", "what imports this file", "impact analysis", "refactoring safety", "review this PR", "review this diff", "is this change risky", breaking changes, unused exports, dead code detection, codemap, file logic, module resolution, graph-it, dependency graph, reverse dependencies.
GraphQL schema design and review. Use when designing new GraphQL schema changes (types, mutations, queries, connections, enums, errors), reviewing schema diffs, planning schema evolution, or auditing nullability and naming. Triggers on 'graphql schema design', 'design a mutation', 'design a query', 'design a type', 'new type', 'new mutation', 'review schema', 'review schema diff', 'schema review', 'schema evolution', 'audit the schema', '/graphql-schema-design'.
Clarify product, domain, scope, constraints, business rules, and technology decisions through dependency-aware question sets. Read the repository first, ask all useful questions whose prerequisites are already known in each set, and use the answers to form later sets. Use when a project or feature is ambiguous and not ready for specification. When decisions are stable, recommend exactly one next skill without invoking it.
Use when reviewing PRs, simplifying over-engineered code, judging architecture, or renaming unclear concepts — inline fake helpers, delete meatless ceremony, rename lying names, merge over-split files, reject premature abstraction, say no to speculative config/modes/layers, then end with the smallest safe next change. Triggers when code is too fancy, too abstract, too clever, too many files/helpers/layers, or too well-factored but painful to change. Embody Grug brain: complexity very bad, small words, no consultant speak, no hard pivot to opposite dogma.
Use when the user asks to “create a hook,” “add a PreToolUse/PostToolUse/Stop hook,” “validate tool use,” “implement prompt-based hooks,” “use ${CLAUDE_PLUGIN_ROOT},” “set up event-driven automation,” “block dangerous commands,” or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks, focusing on the advanced prompt-based hooks API
User-invoked maintenance for openspec/BACKLOG.md. Reconciles deferred-slice entries against the shipped specs in openspec/specs/, proposes deleting entries now covered, and edits the file only on the user's per-entry confirmation. Not part of the tempa-spec pipeline. Run it by hand, ideally right after openspec archive.
Change-size check run inside the tempa-spec schema's `proposal` phase, before the \"What Changes\" list is finalized. Detects when a draft bundles several independent changes that each deserve their own change, stops and asks the user whether to split (default: one change; the agent never splits on its own), and when the user splits, records the deferred slices in proposal.md's Out of Scope section, discovery.md's Key Decisions, and openspec/BACKLOG.md. Rationale is in docs/research/2026-08-27-mattpocock-skills-gate-and-seams.md.
Build, annotate, review, screenshot, and deliver native HTML UI prototypes with AI agents using reusable UI packs, DOM-bound product annotations, in-browser Direct Edit, browser review pins, a local authoring server, an element-to-source Inspector, and multi-state screenshots. Use when the user mentions /html-prototype-build or asks to reconstruct UI in HTML, create product annotations, edit prototype styles, review or mark DOM elements, capture prototype states, package prototype deliverables, or add components to the module sketch board. Works with Claude Code, Codex, Cursor, and other Agent Skills clients; do not use for generic source-code navigation, ordinary documentation, or production frontend development.
Use when reading or editing config code that imports hydra_zen (builds, make_custom_builds_fn, store, just, MISSING, hydrated_dataclass, kwargs_of, make_config, ZenStore, BuildsFn), or when Hydra-style "${...}" interpolations, config groups, _target_ blocks, or `builds_bases` inheritance appear in a hydra-zen project. Covers the typing model, idiomatic patterns, type-refinement gotchas, and hacks for corner cases (partial-of-partial, per-variant overrides, project-wide builder defaults).
Scandit ID Capture (`IdCapture`) in Capacitor projects — scanning passports, driver's licenses, ID cards, residence permits, visas via MRZ, VIZ, barcode, or mobile documents. Use for integration, accepted-document and scanner configuration, captured-field result handling, anonymization, add-on capabilities (voided-ID detection, European driving-license back decoding, AAMVA barcode verification), and SDK version migration.
Automate JIRA ticket processing through domain analysis, initialization, prompt evaluation-optimization and final task execution with comprehensive error handling in thinking model. Use when given a JIRA ticket URL (format: https://[domain].atlassian.net/browse/[TICKET-ID]) to process end-to-end.
Implement exactly one ready Phat task within its documented scope, update its Goal/Blocked by/Todo checklist with verified progress, preserve unrelated changes, apply the project's verification and commit policy, and prepare the task for review. Use when a task is todo or an existing in-progress continuation with all dependencies complete.
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.