Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
Modernize QUnit unit tests and OPA5 integration tests to the UI5 Test Starter concept. Use this skill when: - The linter reports `prefer-test-starter` for *.qunit.html or *.qunit.js files - Test HTML files use manual sap-ui-core.js bootstrapping instead of Test Starter's runTest.js/createSuite.js - Test JS files use Core.ready(), Core.attachInit(), or jsUnitTestSuite instead of sap.ui.define - OPA test HTML files exist with per-test Opa5.extendConfig and manual bootstrapping - An AllJourneys.js orchestrator loads OPA journeys dynamically - OPA journeys call `iStartMyUIComponent` instead of `iStartMyAppInAFrame` - User asks to modernize tests, modernize test infrastructure, or adopt Test Starter Handles unit tests (Core.ready removal, sap.ui.define wrapping) and OPA challenges (page-object imports, Opa5 config, journey orchestration, QUnit 1.x assert modernization, in-window-to-iframe launcher migration). Trigger on: prefer-test-starter warnings, test modernization requests, iStartMyUIComponent.
Gate an implementation plan BEFORE agents execute it — catch phantom paths and false reuse, not typos: read the full plan markdown and its companion JSON, reality-check every file path and reuse claim against the actual repo, challenge scope, architecture, the dependency graph, and risk and test coverage, then score each section and render an APPROVE / REVISE / REJECT verdict — runs as a forked review workflow with its own reasoning budget, isolated from the main execution flow. The verdict follows only from evidence-backed findings: never rubber-stamps a non-trivial plan and never invents findings, but a genuinely clean section is allowed to pass; read-only, never edits the plan. Use when the user asks to review or gate a plan before execution.
Generates a heat-map and metrics report of a repository based on code complexity, lack of tests, and 'TODO/FIXME' density. Use when you need to identify high-risk areas for refactoring or when planning technical debt reduction sprints.
Enforces test doctrine where shape follows code, not the pyramid: pushes logic into Flutter-free packages tested with pure package:test and an injected Clock; asserts invariants with seeded fuzz tests against an independent oracle plus round-trip and rounding goldens; prefers bare-implements fakes over mocktail for code you own; tests the data layer against a real NativeDatabase.memory Drift engine, never a mocked DAO; drives Notifiers headlessly with ProviderContainer overrides; guards one end-to-end acceptance gate and a runtime invariant tripwire; floors coverage on unrecoverable-bug files, not a global percentage; and fixes the coverage-lies-upward gap. Use when writing tests under test/ or integration_test/, choosing unit vs widget vs integration, adding a fuzz or round-trip property, wiring a fake or ProviderContainer test, gating coverage, or triaging a flaky-suite failure.
Run vLLM performance benchmark using synthetic random data to measure throughput, TTFT (Time to First Token), TPOT (Time per Output Token), and other key performance metrics. Use when the user wants to quickly test vLLM serving performance without downloading external datasets.
Expert knowledge for Azure Test Plans development including limits & quotas, security, configuration, and integrations & coding patterns. Use when configuring test failure types, custom fields, retention, permissions, or tcm.exe-based test suite automation, and other Azure Test Plans related development tasks. Not for Azure DevOps (use azure-devops), Azure Pipelines (use azure-pipelines), Azure Boards (use azure-boards), Azure App Testing (use azure-app-testing).
Skill set that enforces design system compliance when Claude writes to Figma through the Figma MCP server: components are instances of master components, and colors/fonts/spacing/radii bind to Figma variables and styles instead of raw values. Every canvas write is followed by a binding check. Needs a Figma file with a design system and the Figma MCP server.
Risk-scaled repo execution and code-evidence protocol. Skip architecture-only work, explanation, contract-preserving prose, and status. Use for contract changes, debugging, code review, implementation plans, and Git mutation; mixed tasks: only those parts.
Investigate the quality of PostHog MCP tool calls — error rates, latency, reach, and which tools are failing or slow. Use when the user asks "which MCP tool has the highest error rate?", "what's the slowest tool?", "which tools fail most often?", "how reliable is tool X?", wants a tool-quality matrix, or pastes an MCP analytics tool-quality / dashboard URL and asks what it shows.
Hunt REAL bugs and security holes in the current branch diff — verified, not a raw model dump: map each changed file's attack and defect surface, run the security-and-logic checklist against it, then verify every candidate against surrounding code, callers, and existing mitigations — runs as a forked review workflow with its own reasoning budget, isolated from the main flow. Reports only findings that survive the false-positive check, severity-ranked with file:line evidence and a fix direction, and states an explicit clean result when nothing real is found; stays read-only and never edits code. Use when the user asks to find bugs, review changes, or audit branch risk.
Analyze any API for AI agent compatibility. Scans OpenAPI specs across 8 pillars (48 checks), scores agent-readiness 0-100, and provides prioritized fix recommendations. Use this skill when the user asks: “Is my API agent-ready?”, “Scan my API”, “Analyze my OpenAPI spec”, “What’s wrong with my API for AI agents?”, “How agent-friendly is my API?”, or wants to improve their API for AI consumption. Works with local specs and Postman collections via MCP.
Review an existing web app and report what is actually wrong with it—security holes checked against the OWASP Top 10, search and AI discoverability that has drifted out of date, and anything the app claims that it no longer does. Use when the user asks to review, audit, or check an app; wants a security review or vulnerability check of a whole codebase rather than a diff; asks whether their sitemap, robots.txt, or llms.txt are still correct; or suspects their landing page, docs, or privacy policy have fallen behind the product. Reviews the app as it stands today, not a pull request or a build sheet, so it catches what was already wrong before the current branch. Read-only by default; gathers evidence once, runs independent lenses over it in parallel, re-checks anything uncertain with a command, and names every check it could not perform rather than implying it passed
Drive the real `10x` CLI (@przeprogramowani/10x-cli) end to end the way a learner does — login, list, get, sync, changelog, doctor — against a local fake delivery API with isolated config and project dirs, and capture stdout/stderr/exit codes, PTY screens and file-tree hashes as proof. Use after changing a command, the writer, conflict handling, auth or the API client, when unit tests pass but you need to see the shipped behaviour, or when asked to "verify", "run the CLI", "prove it works".
Expert knowledge for Chaos Studio development, including troubleshooting, best practices, decision-making, limits and quotas, security, configuration, integrations and coding patterns, and deployment. Use when automating Chaos Studio via CLI/ARM/REST, configuring AKS pod faults, VNet-injected agents, or Workspaces vs. classic, and for other Chaos Studio development tasks. Not for Azure Monitor (use azure-monitor), Azure Resiliency (use azure-resiliency), Azure Reliability (use azure-reliability), or Azure Site Recovery (use azure-site-recovery).
Mandatory review gate for any mesh or asset created or edited in Blender. Detects z-fighting (coplanar overlaps), floating/detached parts, inverted or inconsistent normals, non-manifold/open geometry, degenerate faces, duplicate verts, n-gons, unapplied scale, UV problems, bad skin weights, and renders multi-angle review sheets. Use before calling any modeling, export or delivery task done.
A five-step development workflow. Supports requirements understanding, solution design, implementation, code review, and bug fixing. Use when the user mentions “requirements analysis,” “solution design,” “implementation,” “code review,” “understand the requirements,” “technical design,” “start coding,” “Review,” “check the code,” “bug,” “error,” “crash,” “exception,” or “something went wrong.”
Drive the running app in a real browser to prove a flow works — on demand, or as a saved test for the few flows worth re-running. Use when a multi-step flow needs checking end to end, right after deploying (against the live URL, not localhost), when the builder asks for Playwright, e2e or browser tests, and when something works locally but is reported broken in production. Also use before demo day to confirm the core action still works and to capture screenshots.
Look up an R package's live status on cran.r-project.org - submission/review state (queue, human review, waiting, archived, past version's fate) or R CMD check results (OK/NOTE/WARN/ERROR per platform). Use for "what's the CRAN status of X", "did X get accepted/rejected", "is X passing CRAN checks", "when was version Y archived", "did win-builder get my package", "is win-builder processing my submission".
Enforces a disciplined widget/layout/golden/a11y test surface — one pumpApp harness that pins tester.view.physicalSize x devicePixelRatio and layers MediaQuery (textScaler/boldText/accessibleNavigation) above MaterialApp, an overflow net that never suppresses (one testWidgets per device x scale x bold tuple because overflow reports once per RenderObject), a computed getSize/getRect fit-and-geometry gate instead of goldens, two golden lanes (Ahem geometry + one pinned-OS real-font) with loadAppFonts and blocked --update-goldens, RTL goldens under Directionality, pure-Dart WCAG/APCA contrast on colour VALUES, and honest limits on meetsGuideline. Use when writing test/support/harness.dart, calling pumpWidget/pumpApp, chasing an "overflowed by N pixels" failure, reaching for takeException/ignoreOverflowErrors/FittedBox/withClampedTextScaling, adding matchesGoldenFile, or writing a11y_test.dart with isSemantics/simulatedAccessibilityTraversal/meetsGuideline.
Load this skill immediately whenever a project contains an ACCESSIBILITY.md file, or whenever you are contributing to the mgifford/ACCESSIBILITY.md repository. Absolutely always read ACCESSIBILITY.md before proposing or writing any changes. Under no circumstances skip this skill when an ACCESSIBILITY.md file is present. This skill governs when and how all other accessibility topic skills must be loaded.
Use when designing or reviewing Android UI — applies Material Design 3 UX principles covering touch targets, spacing, navigation patterns, accessibility, animation timing, and platform conventions. Includes an M3 compliance audit that scores screens across 10 categories. Complements the compose skill with design-level decisions.
Maintain profiler performance cases containing only JSONL under ascend-kernel's csrc/ops/<op>/test. Collect data with torch_npu.profiler (fixed warmup=5, active=5), summarize Total Time(us) from ASCEND_PROFILER_OUTPUT/op_statistic.csv, and output a standardized Markdown comparison report (custom operator vs. benchmark) with a DType column. Do not generate perf_cases.json or *_profiler_results.json. See examples/layer_norm_profiler_reference/ for a reference implementation.
Gate a branch before it becomes a PR — catch the structural defects tests miss, not style: read the full diff against the default base, run a two-pass checklist (blocking pass first — unsafe queries, race conditions, trust-boundary and auth mistakes, conditional side effects — then the non-blocking remainder), and verify each candidate against surrounding code before reporting. Stays read-only and reports blocking-first; only becomes mutating after you explicitly approve a specific fix per issue — never batches decisions or auto-fixes. Use as the branch gate right before shipping or opening a PR.
Godot Expert Analyst: Anara. Visionary project scoring and certification for Godot 4.7+ architecture. Use when evaluating project health, modernity, scalability, dependency graphs, or generating Visionary Certificates. Keywords: analyst, Anara, scoring, certification, architecture audit, ResourceLoader.get_dependencies, typed Dictionary, folder-by-feature, Visionary Certificate, Godot 4.7.
Analyze V8 heap snapshots to investigate memory leaks and retention issues. Use when given .heapsnapshot files, asked to compare before/after snapshots, asked to find what retains objects, or investigating why objects survive GC. Provides snapshot parsing, comparison, retainer-path helpers, and scratchpad scripts.
Decide how much proof a piece of work needs — from just looking at it, to walking the flow, to writing a test, to writing the test first. Use right after a feature works, before changing code that already works, and every time something that used to work breaks. Also use when the builder asks whether they need tests, or when a change touches money, auth, permissions, data they can't get back, or logic with rules and edge cases.
ALWAYS use when writing, debugging, or reviewing code that imports “@vue/test-utils” or mentions vue test-utils, VTU, mount, shallowMount, findComponent, wrapper queries, or migrating from Vue Test Utils v1. Provides the current API surface, mounting options, wrapper methods, and testing recipes for @vue/test-utils 2.5.1 on Vue 3
Ascend C code review skill. Reviews code for safety and compliance based on a hypothesis-testing methodology. When invoking, you must explicitly provide: the code snippet and a description of the review rules. TRIGGER when: the user requests code inspection or code review, asks about code security, wants to check coding standards, or needs to investigate a specific code issue (such as a memory leak, integer overflow, or null pointer). Keywords: Ascend C, code inspection, code review, safety and compliance, memory, pointers, overflow, leaks, coding standards.
C/C++/CAPL best practices for automotive embedded systems. This skill should be used when writing, reviewing, or refactoring embedded C/C++ code or CAPL scripts targeting automotive ECUs, following MISRA, AUTOSAR, ISO 26262, and ISO 21434 guidelines. Triggers on tasks involving embedded firmware, CAN/CAN FD/LIN/Ethernet communication, TCP/UDP/DoIP/SOME-IP protocols, RTOS programming, safety-critical code, cybersecurity, diagnostics (UDS), CAPL test automation, or calibration toolchain integration.
Expert knowledge for Playwright Workspaces development including troubleshooting, best practices, decision making, limits & quotas, security, and configuration. Use when managing Playwright Testing workspaces, tokens/RBAC, quotas, monitoring/metrics, or run/AADSTS7000112 issues, and other Playwright Workspaces related development tasks. Not for Azure App Testing (use azure-app-testing), Azure DevOps (use azure-devops), Azure Pipelines (use azure-pipelines), Azure Test Plans (use azure-test-plans).
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-*.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating user timing marks.
Deterministic design-system audit plus a full set of design specialists as slash commands. `npx dwic-audit` scans a project across categories such as color, typography, spacing, accessibility, forms, navigation, motion, and copy using WCAG contrast math and token parsing (no LLM involved), then routes each finding to the specialist that fixes it. Specialists cover visual hierarchy, interaction design, design system architecture, typography, color, spacing, motion, forms, navigation, dashboards, mobile, landing pages, auth UX, healthcare UX, B2B SaaS, e-commerce, dark mode, error handling, onboarding, and data viz. Installs as a Claude Code plugin; the commands are pure markdown with no runtime dependencies.
Review design specification documents for Go programs before implementation begins. Use this skill when a user has a Go spec, design doc, RFC, or PRD to review, wants feedback on a planned feature in a Go codebase, is about to start implementing from a spec, or asks “is this spec ready?” or “review this before I build it”. Applies Go philosophy—simplicity, composition, explicit errors, context propagation—plus Cobra/Viper CLI conventions where applicable.
Use when drafting, saving, formatting, validating, or repairing LinkedIn profile and content assets in this repository, including headlines, About sections, Experience descriptions, project copy, posts, articles, and LinkedIn image specs.
Use when Godot 4.7 work must run without a mouse or at scale: batch import of hundreds of GLB props or an AI image-to-3D art drop, naming rules, import presets and post-import scripts, editor plugins, EditorScript and @tool, EditorImportPlugin, GUT or gdUnit4 tests in CI, a GitHub Actions export workflow, GDExtension basics, or a Godot MCP server; also when --import is slow, textures stay uncompressed, tests pass locally but CI fails, or an export ships test folders.
Creates new skills, modifies and improves existing skills, and measures skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Vitest 3 features and migration from Vitest 2 (2025). Proactively activate for: (1) Vitest 3 stable features, (2) workspace mode for monorepos, (3) test annotations and reporter API, (4) projects configuration, (5) browser mode in Vitest 3, (6) inline snapshots and toMatchInlineSnapshot, (7) typecheck mode (vitest typecheck), (8) watch mode UI improvements, (9) custom reporters, and (10) migration from Jest and Vitest 2.x. Provides: Vitest 3 changelog, workspace setup, projects config templates, browser-mode setup, and migration patterns from Jest and Vitest 2
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.