Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
Use when designing and auditing multiple modules, system-wide quality trade-offs, data authority, target/transition, and migration/recovery from product value. Not for local class improvements, selecting trendy technologies, or extracting a single contract.
Use when converting natural-language requirements into preconditions, postconditions, invariants, and failure guarantees, plus per-condition guarantee responsibility and contract tests. Not for UI input validation only, adding general tests, or whole business implementations.
Use when auditing a fixed use case for missing terms, concepts, value constraints, states, transitions, failures, time, writer/reader, and authority. Not for modeling the whole world, diagrams only, general DDD explanations, or mere data item lists.
Use when separating the purpose, operations, and contracts known to consumers from internal technology and procedures, and designing/auditing technology leakage, caller branching, responsibility boundaries, and change locality. Not for reducing ifs, mass-producing interfaces, or whole architecture selection.
Router used when integrating two or more specialized deliverables, or an end-to-end trace from requirements to implementation, review, and verification. Not for a single design deliverable or small mechanical changes that do not touch public semantics.
Run a Mixture of Experts (MoE) audit on any topic, plan, codebase, evidence archive, or process. Dynamically spawns specialized subagents with different critical lenses to cross-reference findings and detect flaws, overlap, retention issues, or drift. Use when designing architectures, analyzing complex code, verifying multi-step plans, classifying evidence artifacts, or looking for duplicated intent in a repo.
Generate unit test cases for OpenHarmony C/C++ code. Use when: (1) the user asks to generate unit tests for a subsystem/module/file/function; (2) writing HWTEST/HWTEST_F test cases; (3) creating ohos_unittest; (4) the user mentions 'generate test cases', 'write unit tests', HWTEST, ohos_unittest. Keywords: HWTEST, ohos_unittest, unit test, C++ testing, unit test generation
Create, change and debug openQA tests and review openQA results. Use whenever the user mentions openQA, o3, os-autoinst, isotovideo, testapi calls such as assert_script_run or assert_screen, needles, a job URL such as openqa.opensuse.org/tests/123, an openQA job group or build, soft-failures, force_result or verification runs, or asks why an openQA job failed, how to write, schedule, run or fix an openQA test or review a test PR, or to review or label openQA results. Tests: os-autoinst test modules in Perl or Python, YAML schedules, test data, pull requests to os-autoinst-distri-opensuse or a custom test distribution. Results: triage a failed, incomplete or timed-out job from its logs, sweep a job group or build, classify product bug vs test issue vs infrastructure, draft bug reports and job comments whose bugrefs, labels and carry-over openQA parses. Not for OBS package build failures, maintenance-update validation with mtui, plain bug-tracker work, other CI systems or general Perl questions.
A skill that handles E2E scenario generation, execution, and auto-fixing in one place. Triggered by keywords such as "시나리오 만들어줘", "시나리오 실행", "e2e 실행", "E2E 돌려", "시나리오 돌려", "시나리오 변환", "녹화 변환", "e2e 변환", "codegen 변환", "e2e 생성", "시나리오 생성", "시나리오 검증".
Performance issue analysis and optimization workflow, six stages. All triggers start with 性能 ("performance"): 性能分析, 性能证据, 性能定位, 性能假设, 性能监控, 性能优化, 性能验证, 性能深入. Saying one of these words alone, or using the 「trigger: description」 form, enters this workflow or the corresponding stage.
Open-source AI observability platform for LLM tracing, evaluation, and monitoring. Use when debugging LLM applications with detailed traces, running evaluations on datasets, or monitoring production AI systems with real-time insights.
Review a diff (pull request, commit, or local changes) by comparing it against the project's BLUEPRINT.md slice catalog. Auto-detects the input format (git diff, GitHub/GitLab PR URL or number, commit SHA, pasted diff). Loads the blueprint for each project the diff touches and classifies findings as deviations from declared slices, universal anti-patterns, or improvements. Human-in-the-loop on ambiguity (slice classification, severity, ambiguous deviations). Output is ephemeral — emitted to chat or saved on request, never auto-committed. Use whenever the user asks "review this PR", "check my changes", "any anti-patterns in this commit", "is this safe to merge", "look at PR #N", "review my diff", pastes a diff, mentions a commit SHA, or wants a pre-merge sanity check on their code changes.
Apply to multi-step work (sweeps, migrations, runs of similar edits) and to how you stack commits and PRs. Break work into small units that each end in a verifiable state, check each before the next, and order delivery so the sequence proves itself to a reviewer.
Access and test Vercel deployments protected by Vercel Authentication, SSO, or Deployment Protection with agent-browser. Use when a preview or production URL redirects to a Vercel login page, returns a protection 401 or 403, or needs short-lived Trusted Sources OIDC authentication instead of a static bypass secret or public exception.
Systematically QA test a web application. Use when asked to "qa", "QA", "test this site", "find bugs", "dogfood", or review quality. Three modes: full (systematic exploration), quick (30-second smoke test), regression (compare against baseline). Produces structured report with health score, screenshots, and repro steps.
Apply React and Next.js performance optimization guidelines from Vercel Engineering — 40+ rules covering components, data fetching, bundle optimization, and rendering. Triggers on React components, Next.js pages, bundle optimization, performance improvements, refactoring React code.
Advises where a permanent automated test for the Red Hat Developer Hub dynamic-plugin ecosystem should live: which repository (rhdh, rhdh-plugins, or rhdh-plugin-export-overlays), which layer (L1 unit, L2 startTestBackend integration, L3 React Testing Library component, L4a cluster-free Playwright, or L4b cluster e2e), which harness, and which neighbouring file to mirror. Use for "where should this test live", "which repo and layer for this test", "does this need a cluster", "is this e2e too expensive", or a review comment that a test sits at the wrong layer. Advice only — it recommends a placement and names a template; it writes no test and changes no file.
Test authenticated routes in the your project using cookie-based authentication. Use this skill when testing API endpoints, validating route functionality, or debugging authentication issues. Includes patterns for using test-auth-route.js and mock authentication.
Rust code quality with Clippy, rustfmt, and cargo tools. Covers linting, formatting, and idiomatic Rust patterns. USE WHEN: user works with "Rust", "Cargo", "Actix", "Axum", asks about "Clippy", "rustfmt", "Rust linting", "Rust best practices", "idiomatic Rust" DO NOT USE FOR: SonarQube - use `sonarqube` skill, security - use `rust-security` skill
Use when you want to security-test, pentest, audit, or find vulnerabilities in an application you own or are authorized to test — entry point that confirms authorization and routes to the right phase (threat model, code audit, active test, or hardening).
Scaffold and formally review a new Agent Skill in this marketplace repo. Covers valid SKILL.md generation, directory layout, registry entries, and spec auditing. Use when adding a skill, validating frontmatter, or checking marketplace readiness before a PR.
Improves Agent Skills via validate → rule-based eval cases → plugin-eval → prompt evals → bounded edits with held-out gates. Use when tuning skill quality, routing, or adopting Chrome/Microsoft T-named quality gates—not for bulk validate-only or SkillOpt automation.
Analyzes staged and unstaged files from the full working tree, groups them into functional clusters (test, docs, chore, directory prefix), generates a conventional commit message per group, detects common issues (secrets, debug statements, large files), and executes one commit per group sequentially after presenting a multi-commit plan for confirmation. Falls through to a single-commit flow when all detected files resolve to one group. With the opt-in --review flag, first runs a coherence review over the working tree (delegating to /code-review) and surfaces its findings as non-blocking warnings before the commit plan. Trigger: When the user says "commit", "smart commit", or /commit. Add --review (e.g. /commit --review, "commit con review") to enable the coherence gate.
Post-implementation completion workflow for Spec-backed Plans. Use after spec-implement completes to validate, review, create stacked commits, and open a PR via code-pull-request. Triggers only with an active Spec-backed Plan after spec-implement completes, including when the user says "finish", "done", or "complete" in that context.
@DataJpaTest requires TestEntityManager + flush()/clear() before assertions — without this, tests read from Hibernate L1 cache and pass falsely. Read before writing any JPA test. Triggers: @DataJpaTest, JpaRepository, TestEntityManager, @ServiceConnection, Testcontainers, @Modifying, Hibernate 6/7, @AutoConfigureTestDatabase, flush and clear in tests, @Transactional in tests, jakarta.persistence.EntityManager.
Invoke FIRST, before creating, editing, or deleting components, stories, styles, CSS, themes, colors, or design tokens — anything that changes how the UI looks, no exceptions. Also for starting or previewing Storybook to verify UI, requests to show, browse, or list components, stories, or UI states, and docs, props, or usage lookups.
Use this skill when creating comprehensive testing strategies for applications. Provides test planning templates, coverage targets, test case structures, and guidance for unit, integration, E2E, and performance testing. Ensures robust quality assurance across the development lifecycle.
Security-test desktop thick-client applications (.NET/WPF, Java, Electron, native Win32) against their local and network attack surface — proxying non-HTTP traffic, extracting secrets and DB connection strings from config/registry/memory, bypassing client-side trust controls, and reviewing update channels and DLL search order. Use when assessing an installed desktop app that talks to a backend, when Burp sees no traffic from a fat client, or when a two-tier app connects straight to a database.
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.