Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
End-to-end Coval audio-quality testing workflow. Launches one run per audio-robustness scenario against the same voice agent + test set + metrics, polls for completion, and produces the multi-run report URL grouped by Persona. Use when a user wants to follow the Testing Across Audio Qualities cookbook (https://docs.coval.dev/guides/testing-across-audio-qualities) without doing each step by hand.
Assess a reported issue from supplied and bounded read-only evidence before implementation. Use to triage, validate, reproduce, classify, or choose the next step for a report. An explicit target authorizes relevant read-only investigation; provider comments and other mutations require separate authority.
Use when the user wants to measure or set up evals/checks for one of their skills — how fast it is, whether its output is valid, whether it fires when expected, or whether its opening classification/routing gate labels inputs correctly.
Spring Security 7 changed method security defaults and CSRF handling — tests using hasRole() or csrf() may need updates. Read before testing authentication, authorization, or secured endpoints. Triggers: @WithMockUser, csrf(), jwt(), oauth2Login(), SecurityFilterChain, @PreAuthorize, AccessDeniedException, @WithUserDetails, @WithSecurityContext, hasRole, hasAuthority, SCOPE_, JwtAuthenticationToken
Build, review, debug, configure, migrate, or plan Storybook 10 UI workshops with current docs. Use for storybook, create storybook, CSF3, CSF Next, Meta, StoryObj, definePreview, preview.meta, args, argTypes, decorators, parameters, tags, autodocs, play functions, storybook/test, @storybook/react-vite, @storybook/nextjs-vite, @storybook/addon-docs, @storybook/addon-a11y, @storybook/addon-vitest, portable stories, composeStories, main.ts, preview.ts, ESM config, Chromatic, @storybook/addon-mcp, componentsManifest, storybook skills/tools agent CLI, and upgrades from Storybook 8/9 (10.6.x current).
Run the full DEFT AOI improvement loop for NVIDIA TAO VisualChangeNet / ChangeNet PCB inspection models: baseline evaluate, RCA, ingestion of customer-supplied pre-generated AnomalyGen images, k-NN mining, retraining, and deployment gating until FAR / recall KPI targets are met. EA variant — does not run AnomalyGen inline; the customer pre-generates synthetic NG/OK pairs out-of-band and the loop ingests them. Use for prompts like "run the DEFT loop", "fine-tune until FAR below 0.1% at recall=100%", or "improve my AOI ChangeNet model with RCA and pre-generated synthetic defects"; do not use for standalone TAO training, one-off inference, generic anomaly generation, or RCA-only analysis.
Use when auditing tests for TDD smells — phrases like "test smells", "flaky tests", "slow tests", "brittle tests", "test anti-patterns", "tests testing the framework", "over-mocking", "mystery guest", "fragile tests", "test breaks on every refactor". Catalogs the most common test-suite anti-patterns with symptoms, why each is wrong, and the fix.
Use when stuck mid-TDD and unsure which Beck technique applies — phrases like "I have a red test and don't know how to make it green", "should I fake it or just write the code", "is this a triangulation moment", "this test is too big to implement at once", "should I mock this", "what's the next test I should write". Diagnostic mode that routes a stated TDD situation to the relevant tdd-* technique.
Use when substituting a real collaborator with a test double — phrases like "mock object", "test double", "stub", "fake", "spy", "Mockito", "MockK", "should I mock this", "database in tests", "fake the time", "fake the HTTP client". Defines test doubles in their five flavors (dummy / fake / stub / spy / mock) with picking criteria, and the strong warnings about over-mocking.
Use when establishing TDD as a discipline — phrases like "red green refactor", "TDD cycle", "test first", "should I do TDD here", "TDD discipline", "what's the TDD loop", "when should I refactor". Defines the canonical TDD cycle (red → green → refactor), what each phase actually requires, the honest scope of when TDD applies, and the cases where it shouldn't.
Write dialog test specs for a Telegram bot. Covers why tokenless testing exists, how the harness replays Updates and captures API calls, BotSpec JSON format, coverage rules, and the test harness CLI. Tests are the objective publish gate — all specs must pass for the bot to publish. USE FOR: write bot tests, dialog specs, harness, command coverage, BotSpec JSON, tests/specs/*.json, tests/commands/*.json, GATE verdict, coverage gate — even if the user doesn't say "specs" or "test" explicitly. DO NOT USE FOR: programmatic vitest tests with mocks and edge-case fixtures (see telegram-test-advanced), or session persistence (see telegram-bot-sessions). Triggers: write bot tests, dialog specs, harness, command coverage, BotSpec JSON, tests/specs/*.json, tests/commands/*.json, GATE verdict, coverage gate.
Optimize unit, integration and end-to-end tests for type selection, assertions, mocks, test data and review quality. Use when the user asks to add, fix, refactor or review tests, or deals with meaningless assertions, mocks, coverage, spec, integration or e2e boundaries.
Govern flaky tests and suite reliability: rerun-pass verdicts, root-cause classes, quarantine, retry semantics, health metrics. Not for: in-run failures (e2e/api), triage, confirmed bugs. 治理 flaky 测试与套件可靠性：时好时坏/重跑变绿判定、根因四分类、隔离门禁、重试诚实语义、健康度。不用于：执行中单条失败（e2e/api）、批量分流（triage）、已确认 Bug（bug-analysis）。
Use when creating or editing skills, before deployment, to verify they work under pressure and resist rationalization. Applies the RED-GREEN-REFACTOR cycle to process documentation by running a baseline without the skill, writing to address failures, and iterating to close loopholes.
Add a new cuTile GPU kernel operator to TileGym. Covers dispatch registration in ops.py, cuTile backend implementation, __init__.py exports, test creation, and benchmark in tests/benchmark. Use when adding, creating, or implementing a new cuTile operator/kernel in TileGym, or when asking how to register a new cuTile op.
Deliver the verdict for a unit of work once the TESTER and DEV have both submitted: run the TESTER's tests against the DEV's code, arbitrate each failure against the contract (the code deviated, the test deviated, or the contract did not specify it), verify that each role touched only its own paths, and produce the traceability matrix for the PR. Use when the LEAD has received both deliveries (ENTREGA TESTER and ENTREGA DEV) for a unit of work, or when asked to verify, judge, arbitrate, or give the verdict on a team delivery
Use before claiming that work is complete, fixed, or tests have passed, and before committing or creating a PR. You must run verification commands and confirm their output before claiming success; always support claims with evidence.
Use to reconcile CONTEXT.md files with the current state of the code. Detects drift using provenance stamps, classifies divergences by seam firmness, auto-updates soft prose to match code, and surfaces firm divergences for human adjudication. Use after hotfixes, drive-by PRs, or periodically to keep docs honest. Also the tool that runs during implement's docs reconciliation phase.
Use when planning or implementing Vue 3 projects - helps architect component structure, plan feature implementation, and enforce TypeScript-first patterns with Composition API, defineModel for bindings, Testing Library for user-behavior tests, and MSW for API mocking. Especially useful in planning phase to guide proper patterns before writing code.
Testing patterns for Zod schemas using Jest and Vitest. Covers schema correctness testing, mock data generation, error assertion patterns, integration testing with API handlers and forms, snapshot testing with z.toJSONSchema(), and property-based testing. Baseline: zod ^4.6.0. Triggers on: test files for Zod schemas, zod-schema-faker imports, mentions of "test schema", "schema test", "zod mock", "zod test", or schema testing patterns.
Evidence-bounded red-team review that challenges a plan or spec without manufacturing findings. Returns a priority-sorted finding list (P0–P3) with an explicit PASS/FAIL verdict. Use for plans/specs; use risk-calibrated-pr-review for an opened implementation PR, or adversarial-diff-review for an explicitly requested branch/worktree diff.
Orchestrate complex work via a phase-gated multi-agent loop (audit → design → implement → review → validate → deliver). Use when you need to split work into subsystems, run independent audits, reconcile findings into a confirmed issue list, delegate fixes in clusters, enforce PASS/FAIL review gates, and drive an end-to-end validated delivery. Do not use for small, single-file tasks.
Use this skill when reviewing AI-generated unit, functional, API, or end-to-end tests for false confidence, weak assertions, missing risks, or unsafe test behavior; triggers include AI 生成测试评审, AI-generated test review, and functional test review
Plan and audit AI safety red-team reviews for prompt injection, jailbreaks, data exfiltration, tool misuse, unsafe autonomy, hallucinated actions, privacy leakage, harmful content, policy bypass, model routing fallback, eval coverage, mitigations, release gates, and incident response. Use when an AI feature needs adversarial testing before or after launch.
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.