Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
Configure LangWatch online evaluations and guardrails for production traffic. Use when the user wants to score live traces or threads, monitor production quality, sample incoming traffic, or synchronously block unsafe requests and responses. Do not use for batch experiments.
Optimize software performance through user-journey baselines, profiling, representative benchmarks, controlled experiments, and field verification. Use when asked to make an application, service, workflow, or hot path faster or more efficient, or to build the measurement loop first. Do not use for a reliability incident, capacity forecast, generic product A/B test, or operating a named telemetry or load-testing tool without an optimization objective.
Use when the user has a prompt that feeds a system they can already score, and wants that prompt automatically improved to raise the score against their own evaluation command. Makes one targeted quality edit per iteration — clarity, context, specificity, structure, examples, decomposition, guardrails — re-runs the user's eval to measure the metric, and keeps the edit only if the metric improves, else reverts; loops to a target, plateau, or budget. The metric is whatever the user's eval command prints (task accuracy, an LLM-judge score, a pass rate, a tool-call success rate); the loop is metric-agnostic and never edits the eval. Not for writing a prompt from scratch, not for tuning model weights or hyperparameters, and not for one-off manual prompt edits without a score.
Senior Quality Manager Responsible Person (QMR) for HealthTech and MedTech companies. Provides overall quality system responsibility, regulatory compliance oversight, management accountability, and strategic quality leadership. Use for quality system governance, regulatory compliance oversight, management responsibility, and quality strategic planning.
Run a bounded red-teaming autoresearch loop to generate LLM guardrail training data. You (the agent running the skill) are the attacker and the judge: you craft attacks and label every response. The only model the harness calls is the target under test, over any OpenAI-compatible API (OpenRouter, Moonshot/Kimi, Fireworks, Ubicloud, OpenAI, or custom), with its key in `.red-team/.env`. Every attempt (pass and fail) is written to JSONL ready for fine-tuning guardrails in Llama Guard format. Use when asked to red-team or stress-test an LLM for harmful content, jailbreaks, prompt injection, or backdoor/trigger behavior; to build a safe/unsafe dataset; or to mine novel attacks at scale. Complements the hacker skill (code/infra exploitability): this skill targets model behavior and produces data.
Use this skill to find and remove dead code, unused dependencies, duplicate logic, and unused exports using detection tools (knip, depcheck, ts-prune, deadcode, staticcheck). Trigger on "clean up dead code", "remove unused", "find dead code", "reduce bundle size", "dependency audit", or when the codebase feels bloated. For simplifying living code (readability, naming, complexity reduction) without detection tools, use code-simplifier instead. Use this skill during maintenance windows or before major refactors.
Audit code for DRY violations, dead code, complexity, and consistency issues. Read-only analysis with actionable recommendations. Use before PR or for code quality review. Triggers: review maintainability, code quality, DRY, refactor review.
Use after run-plan completes to independently verify the implementation. Reads only the plan document and inspects the codebase from scratch — information-isolated from the execution context. Produces a structured review document with PASS/FAIL verdict. Triggers when the user says "review the work", "verify the implementation", "check if the plan was executed correctly".
Enforcing the RED-GREEN-REFACTOR loop: write one failing test and watch it fail, write minimal code to pass, then refactor green. Use when starting implementation of a new feature or bugfix, or writing any new production code. Requires pasted failure output as proof of RED; code written before its test must be deleted, not stashed. Skip for exploratory spikes or when only modifying existing tests.
ALWAYS USE before running .NET tests or answering with a test command or flags. Trigger on "run the tests", "exact dotnet test command", one test/class/category/trait/target framework, combined filters, `--filter-query`, `--no-build`, `--diag`, diagnostic logs, TRX, coverage collection, crash/hang dumps, filter errors, or unrecognized options. Chooses repository-compatible classic, VSTest, bridged MTP, or native MTP syntax for MSTest/xUnit/NUnit/TUnit. DO NOT USE for platform identification alone (platform-detection), writing or debugging test code, interpreting an existing coverage report, CI investigation, migration, or a persistent hot reload/watch loop.
Use when the user asks for a second-opinion review from another agent (e.g., "get a second opinion", "review with pi", "review using gpt-6-astra", "/second-opinion"), or when review-code runs with `--second-opinion`. Runs a fresh pi session (or Claude Code, when already inside pi) on the review-code skill and relays its review.
CI-only self-healing workflow using gh-aw (GitHub Agentic Workflows) for active runtime recovery on pull requests and scheduled runs. When a CI check fails (test, build, lint, deploy, scan), this skill diagnoses the failure from CI logs, proposes a verified patch as a PR comment or follow-up commit, and commits a HEAL entry to `.learnings/HEALS.md`. Verify-before-persist discipline preserved: a HEAL is only `verified` if a re-run check passes in the same workflow; otherwise it ships as `pending-verify` for human follow-up. Recurrent heal patterns across PRs accumulate `Recurrence-Count` and append a `Handoff` block at ≥3 to flag promotion via self-improvement-ci. Use this skill when: you want headless heal-loop execution in CI/scheduled pipelines, you want recurring failure patterns captured automatically, or you want PRs that surface non-obvious environmental / tooling fixes without human triage. For interactive/local sessions, use `self-healing` instead.
Set up the `clickhouse-js` repository in a fresh checkout so the agent can run tests, lints, type checks, builds, or examples. Use this skill before invoking any `npm run test:*`, `npm run lint`, `npm run typecheck`, `npm run build`, or `npm run run-examples` script — or after pulling changes that touch any `package.json` (root, `examples/node`, or `examples/web`). Covers Node.js version requirements, installing dependencies across the npm workspaces and the two independent example packages, building the workspace packages so inter-package imports resolve, and starting ClickHouse via Docker Compose for integration tests. Do NOT use this skill for downstream user projects that merely depend on `@clickhouse/client` or `@clickhouse/client-web`; it is specific to contributing to the `ClickHouse/clickhouse-js` repo itself.
Prepare a feature branch for a pull request: verify scope, sync when needed, test, review, push and open or update the PR. This skill MUST be used when the user requests this branch-to-PR workflow or explicitly invokes ship-it (開 PR、準備 PR、整理分支交付). It SHOULD be used when the user asks to prepare a completed feature for review. It MUST NOT be triggered solely by a push, merge, release, deployment request or a general readiness question.
Specification quality checker for spec-generator documents. Validates requirement.md, design.md, tasks.md for consistency, completeness, and quality. Detects requirement ID mismatches, missing sections, contradictions, and ambiguous expressions. English triggers: "inspect specs", "check specification quality", "validate requirements" 日本語トリガー: 「仕様書を検査」「品質チェック」「仕様を検証」「spec-inspect実行」
Boot 4+ replaced MockMvc assertions with MockMvcTester — incompatible API your training data gets wrong. Read before writing any @WebMvcTest. Triggers: @WebMvcTest, MockMvc, MockMvcTester, @RestController, jsonPath, @GetMapping, @PostMapping, @AutoConfigureMockMvc, MockMvcRequestBuilders, status().isOk(), @RestControllerAdvice, contentType(APPLICATION_JSON).
Expert setup assistant for the Stably Playwright SDK. Use this skill when installing Stably SDK in a new project, migrating from @playwright/test, or configuring Stably reporter for CI/CD. Triggers on tasks like "setup stably", "install stably sdk", or "configure playwright with stably".
Find missing interface and service states such as loading, empty, partial, permission, timeout, stale data, validation, failure, interruption, and recovery. Use when a designer asks what states or edge cases are missing from a screen, component, journey, or feature.
Design and audit customer support quality assurance, ticket review scorecards, macro quality, escalation correctness, refund/support fairness, tone, evidence handling, coaching loops, calibration, automation QA, and support health metrics. Use when improving support operations for SaaS, apps, games, marketplaces, developer tools, billing, trust, or community products.
Reviews Svelte components for idiomatic patterns, identifies anti-patterns carried over from React/Vue/Svelte 4, and suggests Svelte 5 refactorings. Use when writing or reviewing Svelte components, asking "is this idiomatic", "Svelte way", doing Svelte 5 migration, or encountering smells like: effect assigns state, prop copied to $state, global store by default, $bindable everywhere, clickable div, createEventDispatcher, export let, on:click, slot-shaped APIs, lifecycle-driven code, imperative DOM wiring, immutable-update ceremony, context value replacement, shadcn-svelte form structure, Field.* components, bits-ui form controls, or component APIs that hide ownership. This is the general-purpose entry point for Svelte component review; delegates to sveltekit for routes/load/actions/server concerns and to focused references for details.
Review and address code changes in Tenzir projects. Use when auditing diffs or pull requests for bugs, security issues, missing tests, documentation drift, readability problems, performance regressions, user experience issues, or when handling pull request review comments, especially implementing requested changes, replying to reviewers, resolving GitHub review threads, or verifying that review feedback has a final state. Also use this skill whenever the user says "review", "look at this PR", "check my changes", "audit this diff", "address the review findings", or "resolve PR comments".
Comprehensive visual regression testing using Playwright and jest-image-snapshot. Implements screenshot comparison, baseline management, CI/CD integration, and visual diff reporting following Ant Design best practices. Use for preventing visual bugs, ensuring UI consistency, and automating visual QA. (project)
Build, review, debug, configure, migrate, teach, or plan Vitest testing with current docs and a full usage guide. Use for vitest 5, vitest.config, defineConfig from vitest/config, describe/it/test/expect, vi mocks, vi.when, snapshots, coverage (@vitest/coverage-v8), browser mode (@vitest/browser-playwright), projects, pools/maxWorkers, reporters, UI, bench fixture, typecheck, vitest doctor, fsModuleCache, and Jest / Vitest 3→4 / 4→5 migration.
Audit a fixed cohort of inactive WooCommerce pull requests against current trunk, linked issues, review history, repository labels, and representative user flows, then produce evidence-backed actions and a private checklist report. Use for stale or backlog PR cohort reviews, not ordinary single-PR code review.
Hard-gate writing auditor. Use to independently audit the hard gates of a writing workflow before formally outputting the final article. Trigger scenarios: - a draft already exists and you are about to produce the formal final version - the project is located at `01_项目/内容创作/` - you need to verify artifacts such as Claim Ledger, Module Budget, Lint Report, Final Candidate Core capabilities: - independently check whether artifacts are complete - check whether form hard thresholds are exceeded - check whether claims match the body text - output `PASS / FAIL` and write it to `03-Reviews/Gate-Check-Report.md`
Executes QA test cases created by /aif-qa in human-guided or automated-agent mode, including reusable browser replay scripts. Use when you need to walk through QA one case at a time, record pass/fail results, or have an agent verify and rerun cases through browser, CLI, API, automated tests, or file/document checks.
Stand up a NEW Bun/TypeScript, Next.js, or Java (Quarkus) repo to the atelier standard from zero — scaffold the Clean Architecture (or Atomic Design) layout, copy the gate assets, wire the git hooks, write the build scripts, lay a minimal green walking skeleton, and prove every gate passes before the first commit. Use when starting a fresh repo or a new monorepo package from scratch — say "scaffold a new Bun repo", "bootstrap a new project to the standard", "set up a new atelier repo / Next.js package", or "scaffold a new Java/Quarkus service". Greenfield only. NEVER use it for work inside an existing codebase (adding features, setting up lint or tooling, config changes) — those belong to the main atelier skill, or to atelier-review-me adopt mode for bringing a brownfield repo up to the standard.
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.