Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
Plan and, after explicit approval, execute runtime-behavior probes for local or live integrations. Use only when explicitly invoked to verify behavior that code review and normal tests cannot settle; define a controlled validation matrix, keep disposable JS probes outside tracked paths, and report observed evidence.
Provides comprehensive guidance for Selenium WebDriver including browser automation, element location, waits, and test frameworks. Use when the user asks about Selenium, needs to automate web browsers, write Selenium tests, or work with Selenium WebDriver.
Investigate, optimize, or review software performance using representative measurements, profiles, budgets, and regression evidence. Use when latency, throughput, resource use, loading, rendering, bundle cost, or scalability is a primary concern. Do not use for speculative micro-optimization without a symptom, target, or demonstrated bottleneck.
Audit and improve SwiftUI runtime performance through code review and architecture. Use to diagnose slow rendering, janky scrolling, high CPU or memory usage, excessive view updates, or layout thrash in SwiftUI apps, and to provide guidance for user-run Instruments profiling when code review alone is insufficient.
Use before implementing a feature or bug fix that adds or changes behavior users or callers rely on, including requests that do not mention tests or TDD. Not for removing features or code, behavior-preserving refactors, design or UX exploration, prototypes, styling, copy, configuration, or data cleanup.
Build or improve UI test coverage for visual regression, interactions, and accessibility. Use when the user asks for a test strategy, automated tests, testing setup, or fixes to flaky UI tests. A one-time browser inspection or visual review does not require a new test suite.
Review currently staged changes along two axes: Standards (does the code follow this repo's documented coding standards?) and Spec (does the code match what the originating task/spec asked for?). Runs both reviews in parallel sub-agents and reports them side by side
Audit existing UI screens, URLs, screenshots, Figma frames, HTML/CSS/React code, dashboards, mobile flows, and landing pages for UX, accessibility, responsive behavior, visual hierarchy, conversion blockers, state coverage, and regression risk. Use this skill whenever the user asks for a UI/UX review, severity matrix, evidence-backed findings, before/after check, or actionable refactor brief. Never claim DOM, Lighthouse, axe, or screenshot evidence unless it exists.
React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.
Check that a form actually rejects bad input: the error message renders, the submit button stays disabled, and no request goes out. Use when validation logic was written but never driven, when a "required" or pattern check looks right on screen but was never proven, or when a bug report says the form submitted invalid data anyway.
Use this skill whenever you need to verify that a web UI actually works the way it's supposed to — clicking through flows, asserting state, catching console errors, taking screenshots across multiple viewports for visual review. Triggers on phrases like "驗證網頁", "verify the site", "Playwright tests", "visual regression", "RWD verification", "screenshot comparison", "responsive check", "console error check", "看看手機版有沒有壞", "視覺驗證", "Playwright 測試", "截圖比對", "RWD 驗證", "驗 sidebar", "看 console 有沒有錯", or any post-change moment where the user wants to know "did I break anything?". This is the runtime-behaviour verifier — use `web-content-audit` instead if the question is about file/data consistency rather than rendered behaviour.
Reviews Workers and Cloudflare Developer Platform code for type correctness, API usage, and configuration validity. Load when reviewing TypeScript/JavaScript using Workers APIs, wrangler.jsonc/toml config, or Cloudflare bindings (KV, R2, D1, Durable Objects, Queues, Vectorize, AI, Hyperdrive).
Headless WordPress and WPGraphQL review guidance. Use when reviewing decoupled WordPress architectures, WPGraphQL schema design, `register_graphql_field`, `register_graphql_connection`, `graphql_register_types`, Next.js/Gatsby/Remix frontends, preview mode, auth flows, persisted queries, or build/revalidation pipelines. Helps review schema boundaries, resolver performance, preview/auth correctness, cache invalidation, and content modeling decisions for headless WordPress stacks.
Use this skill to write tests with the `kotlin.test` library — the multiplatform assertion + annotation API that compiles the same in `commonTest` and on the JVM (JUnit4 or JUnit5), Android, JS, Native, and Wasm. Covers the `kotlin.test` annotations (`@Test`, `@Ignore`, `@BeforeTest`, `@AfterTest`) and how they typealias onto the underlying framework; the `assert*` functions (`assertEquals`/`assertNotEquals`, `assertSame`, `assertNull`/`assertNotNull`, `assertIs`/`assertIsNot`, `assertContentEquals`, `assertContains`, `assertTrue`/`assertFalse`, `fail`, `expect`, `assertFails`, `assertFailsWith`); the `@OnlyInputTypes` compile-time type check; the `Asserter`/`AsserterContributor`/`DefaultAsserter` extension point that routes JVM failures to `org.junit.Assert` for clickable diffs; and the `kotlin("test")` capability that auto-selects JUnit4 vs JUnit5. Use when the user mentions `kotlin.test`, `import kotlin.test.*`, `@BeforeTest`, `assertFailsWith`, `assertContentEquals`, `assertIs`, or `kotlin("test")`.
ZenTao bug management - query bug lists, view bug details, create bugs, update bug status, resolve bugs, and close bugs. Triggered when the user says: "查询 bug", "有哪些 bug", "bug 列表", "查看 bug", "创建缺陷", "报告 bug", "登记 bug", "缺陷", "bug 数量", "待处理 bug", "bug 状态", "禅道 bug".
Systematically audit and remediate accessibility issues in UI, focusing on keyboard navigation, screen reader support, color contrast, semantic HTML, ARIA usage, and motion sensitivity. Use when the user wants to improve accessibility, make a component accessible, fix WCAG violations, add keyboard support, or ensure screen reader compatibility.
Checks that every Python code block in a Markdown file actually compiles and runs, by extracting each block to a temporary file, executing it in an isolated subprocess, and writing a pass/fail report with per-snippet coverage. Use when the user asks to verify, test, or validate the code samples in a README, a guide, or a documentation page; wants to know which snippets in a Markdown file are broken or out of date; or asks for a snippet verification report. Don't use for running the project's test suite (run pytest directly), for checking code style or formatting (use `adk-style`), or for authoring a new runnable sample agent (use `adk-sample-creator`).
Load this skill when the user wants to run an online experiment (live-traffic A/B test) on an LLM application instrumented with Agent Observability: compare two versions of a prompt, model, or behavior using a Datadog feature flag, score metrics from evaluations, and cost and token metrics. Covers creating the flag, wiring the app, creating the experiment, attaching metrics, setting the traffic split, and starting it. Triggers: online experiment, A/B test an LLM app, test a prompt change on live traffic, experiment with feature flag and evaluation score
Strip comments that do not earn their place. Spawns the Comment Reaper on the caller's files or the current diff, applies the accepted deletions, and fixes or reports the code the comments were covering for. Use when asked to banish comments, remove comments, clean up comments, decomment, or run /banish.
CLI + Backend integration testing workflow. Use when verifying backend API changes (TRPC routers, services, models) via the LobeHub CLI against a local dev server. Triggers on 'cli test', 'test with cli', 'verify with cli', 'local cli test', 'backend test with cli', or when needing to validate server-side changes end-to-end.
Review one GitLab merge request or local WIP for concrete risks and proportionate fixes, preserving agreed scope and prior decisions in follow-ups. Include GitLab labels, SemVer, and manual publication commands for remote reviews. Use when the user asks to review a GitLab merge request or local work-in-progress changes. Russian discovery terms: ревью кода.
Author, validate, and enforce a CODEOWNERS file: detect an existing file, gather ownership evidence from directory structure and git history, write or update rules with correct CODEOWNERS syntax, validate the file against the GitHub API, and wire the ruleset rule that requires code owner review on pull requests. Use when the user says "add CODEOWNERS", "who owns this path", "require owner review", "validate CODEOWNERS", or invokes /codeowners. Idempotent — safe to re-run; diffs before overwriting and never auto-assigns an owner without confirmation. Do NOT use for branch rulesets in general (use /gh-bootstrap) or community health files such as README, LICENSE, CODE_OF_CONDUCT, and SECURITY (use /oss-hygiene) — this skill only covers the CODEOWNERS file and the review rule it feeds.
Same-epoch comparison of training runs across wandb, neptune, tensorboard, or mlflow. Aligns runs at the student's current step (never current-vs-final-of-baseline) and separates proxy metrics from downstream targets. Use when the user asks to compare runs, check if a run is improving, track lag against a baseline, rank experiments, or evaluate run-vs-run performance.
Existing-code comprehension and local winnowing preflight. Use for simplify/refactor/clean up/untangle, nested branches, boolean soup, opaque names, mixed responsibilities, cross-file state, or review stalls. Factor the local whole, separate essential/incidental/specification-risk factors, winnow dominated or duplicated factors, and emit the smallest clarity cut. Not for broad architectural layer removal, kernel quotienting, invariant remediation, or greenfield planning.
Provides comprehensive guidance for Cypress end-to-end testing including commands, assertions, component testing, CI/CD integration, and best practices. Use when the user asks about Cypress, needs to write E2E tests, component tests, or configure Cypress for testing.
Evidence-before-action diagnosis of failing ML experiments. Probes the system before guessing causes, process list, dmesg, GPU stats, log scrollback, checkpoint state, then states a hypothesis as a hypothesis and runs a smoke before claiming a root cause. Use when the user asks why a run is failing, diverging, OOMing, hanging, slow, producing weird metrics, has crashed, or asks to debug, diagnose, troubleshoot, or investigate a training issue.
Systematic debugging methodology with root cause analysis. Phases: investigate, hypothesize, validate, verify. Capabilities: backward call stack tracing, multi-layer validation, verification protocols, symptom analysis, regression prevention. Actions: debug, investigate, trace, analyze, validate, verify bugs. Keywords: debugging, root cause, bug fix, stack trace, error investigation, test failure, exception handling, breakpoint, logging, reproduce, isolate, regression, call stack, symptom vs cause, hypothesis testing, validation, verification protocol. Use when: encountering bugs, analyzing test failures, tracing unexpected behavior, investigating performance issues, preventing regressions, validating fixes before completion claims.
Critique UI visual quality with observable design criteria. Use this skill whenever the user asks for visual critique, design review, polish, hierarchy, spacing, typography, contrast risk, consistency, brand fit, anti-slop review, screenshot critique, Figma frame critique, or a fix brief for a weak interface. Keep findings actionable and evidence-labeled. Do not require Attention Insight, Khroma, or external screenshot upload for normal local critique.
Repair persisted fix-now findings in compatible atomic batches: root-cause fixes, green gate, commit/push, and per-row `folded: yes` updates. Never reclassify or substitute backlog notes. Triggers: "fold-findings", "fix the review findings", "repair audit blockers".
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.