Skills
96,770 packaged capabilities and playbooks for agents. Install any of them with one command.
Testing & quality
Request or submit an approval review on a GrowthBook feature flag draft revision. Use when the user says "request review for this draft", "approve this change", "reject this draft", "request changes on revision X", "I want to review flag Y's pending draft", "submit my approval", "mark this as needing changes", "who needs to approve this", or "check the review status". For creating and editing drafts, use the relevant flag-* write skill. For publishing an approved draft or resolving merge conflicts, use flag-publish. For listing all pending drafts across flags, use flag-revisions.
Review React, Next.js App Router, TypeScript, Tailwind CSS, shadcn/ui, TanStack Query, and frontend UI code. Use when reviewing a PR, branch, diff, page, component, hook, Server Action, route handler, or API client; debugging frontend behavior; assessing production readiness; or checking runtime bugs, client/server boundaries, data fetching, security, accessibility, UX states, responsive layout, performance, SEO metadata, i18n, and maintainability.
Automates GitHub repository setup: 12 production-tested GitHub Actions workflow templates (React/Vite, Node matrix testing, Python, Cloudflare Workers deploy, basic CI), 4 YAML issue and pull request templates, Dependabot config, CodeQL security scanning, branch protection, and CODEOWNERS. Prevents 18 documented errors in workflow YAML syntax and indentation, action version pinning, runner versions, issue template structure, Dependabot PR limits, and CodeQL permissions such as security-events: write. Use when bootstrapping a new repo, adding CI/CD pipelines, creating issue or PR templates, enabling Dependabot or CodeQL scanning, deploying to Cloudflare Workers from GitHub Actions, or fixing GitHub Actions workflow YAML syntax errors.
Answer questions about Glubean docs, help users onboard into a real Glubean setup, derive meaningful tests or contracts from existing implementation, or write/run/fix verification inside a Glubean project. Use when the user asks to learn Glubean, set it up, verify existing code, improve coverage, or work with @glubean/sdk or @glubean/browser, including unpacked Chrome extensions and native Side Panel verification.
GraphQL schema design and review. Use when designing new GraphQL schema changes (types, mutations, queries, connections, enums, errors), reviewing schema diffs, planning schema evolution, or auditing nullability and naming. Triggers on 'graphql schema design', 'design a mutation', 'design a query', 'design a type', 'new type', 'new mutation', 'review schema', 'review schema diff', 'schema review', 'schema evolution', 'audit the schema', '/graphql-schema-design'.
Systematic usability evaluation using established heuristics (Nielsen's 10, Shneiderman's 8, or custom rubrics). Use when reviewing UI designs, screenshots, prototypes, or live products for usability issues. Triggers on "review this design", "what's wrong with this UI", "usability check", "evaluate this interface", or when user shares screenshots/mockups asking for feedback.
Show Honey's benchmark scoreboard — the committed quality and token results per task tier (code, user-facing, agent-to-agent) from bench/. Reports only the reproducible committed figures, never invents per-repo numbers. Use when asked how much Honey saves, how it compares to Caveman / Ponytail / no-skill baseline, or for the headline numbers.
Translates game design prose into engine-neutral implementation invariants and validation checks. Use when implementing or reviewing game mechanics where idle, hold-only, mashing, spam, safe waiting, or repeated scoring pulses must be prevented from dominating skilled play.
Add, configure, and manage faults on Simulink, Simscape, and System Composer model signals for robustness analysis and safety validation. Use when injecting faults (stuck, noise, gain, offset) onto block inports/outports, enabling fault simulation, or analyzing fault effects. Covers sensor failures, signal corruptions, actuator faults, FMEA validation, and robustness testing. Requires Simulink Fault Analyzer.
Light 科研主线 stage 6: turn the frozen question/estimand, experiment matrix, pre-registration, and data lineage into minimal runnable, test-first, leakage-free, reproducible experiment code that can be handed off to result-analysis. Use to implement or reproduce training/preprocessing/evaluation; design gold/property/metamorphic tests; control Python/NumPy/framework/CUDA/DataLoader randomness; review train/test or CV fit leakage; record config/code/environment/input hashes, stdout/stderr, raw metrics, patient/entity predictions, and failure artifacts; and run the stage-6 checkpoint. Data leakage or irreproducibility is critical; static scanning and two runs with the same seed do not prove absolute reproducibility across hardware.
Create test users and add funds to them for Mercado Pago testing. Wraps create_test_user and add_money_test_user from the MCP. Clarifies that credentials come in APP_USR- (Orders API, Checkout Pro, Point, QR) and TEST- (Checkout API, Bricks, Payments API) formats — both are valid and actively issued.
A structured migration workflow for moving native mini program projects to `weapp-vite + wevu + Vue SFC`, covering phased rollout, behavior-equivalent conversion, typed props/events migration, platform guards, plus migration-oriented e2e verification and rollback checkpoints. Use for scenarios such as "migrating native to weapp-vite", "setData refactoring", "migration rollback strategy", or "how to test migration e2e".
Drive a PR or branch to green CI. Reads check status through tracker operations, pulls failed-step logs, classifies each failure (real bug, test bug, flake, infra), fixes the real ones in an isolated worktree with tests, pushes, and re-checks — iterating until every required check passes or a genuine blocker is reported. Never goes green by weakening tests or disabling checks.
Use this skill when you need to design JMeter test plans with Thread Groups, samplers, data sets, assertions, timers, CLI runs, and HTML reports; triggers include JMeter performance testing, performance testing, and performance-test-jmeter.
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
Use this skill when project delivery constraints or action tracking are needed for test strategy, test strategy review, or test report review; triggers include project delivery perspective, delivery planning, schedule and capacity.
Use when managing prompts in production at scale: versioning prompts, running A/B tests on prompts, building prompt registries, preventing prompt regressions, or creating eval pipelines for production AI features. Triggers: 'manage prompts in production', 'prompt versioning', 'prompt regression', 'prompt A/B test', 'prompt registry', 'eval pipeline'. NOT for writing or improving individual prompts (use senior-prompt-engineer). NOT for RAG pipeline design (use rag-architect). NOT for LLM cost reduction (use llm-cost-optimizer).
Rails 8.x application architecture, implementation, and review guidance for production codebases. Use when building or reviewing Ruby on Rails 8 features across models, controllers, routes, Hotwire, jobs, APIs, performance, security, and testing. Trigger for requests mentioning Rails 8, Active Record, Active Job, GoodJob, Solid Queue, Turbo/Stimulus, REST resources, migrations, code quality, naming, and production readiness.
Performs security audits and vulnerability assessments on Ruby on Rails application code. Use when reviewing Rails code for security risks, assessing authentication or authorization, auditing parameter handling, redirects, file uploads, secrets management, or checking for XSS, CSRF, SSRF, SQL injection, and other common vulnerabilities.
Use when changing existing structure without changing behavior, adding characterization tests, isolating hard dependencies, consolidating APIs, migrating callers, introducing value objects or domain events, replacing null/string/generic exceptions with typed failure contracts, or removing legacy paths safely. Applies to protected incremental change, not an unverified rewrite or unrelated cleanup.
Improve the structure of working code without changing what it does: extract duplication, break up complex functions, remove dead code, replace magic numbers, fix misleading names, and modernize to the language's idioms — each step verified against the tests. Detects the stack from its build files and leans on the project's own linter (detekt/ktlint, dart analyze, SwiftLint, cargo clippy, clang-tidy). Use when someone says 'refactor this', 'clean this up', 'this function is too long', 'there's a lot of copy-paste here', 'remove the dead code', 'simplify this', 'make this more idiomatic', 'reduce the complexity', or the Hungarian 'refaktoráld ezt', 'tisztítsd meg ezt a kódot', 'túl bonyolult ez a függvény', 'sok itt a duplikáció'. This is the apply counterpart to code-analyzer's detect: that skill finds what is wrong across a project, this one changes it. Not for fixing a bug or crash (that is error-debugging), nor for writing the tests a refactoring needs (that is test-generation).
Design and audit reliability incident learning, postmortems, incident timelines, contributing factors, action items, recurrence prevention, customer impact, status updates, error budget learning, operational follow-through, and blameless review. Use when outages, degraded service, data incidents, support spikes, or release failures need durable improvement instead of shallow postmortems.
Reviews Remix v2 code for caching header misuse, missing server/client split, hydration mismatches (Date, Math.random, locale), prefetch hygiene, and asset bottlenecks. Use when reviewing routes that export headers, use .server.ts/.client.ts, or render dates/IDs in a Remix v2 codebase.
Audit two or more interdependent repositories for drift, duplication, and practices one repository uses well and another lacks. Verify every finding mechanically, decide with the user, then apply surgical alignment fixes and adapted transfers per repository.
Comprehensive code review with parallel specialist sub-agents. Analyzes requirements traceability, code quality, security, performance, accessibility, test coverage, and technical debt. Produces detailed findings and calls /qa-gate for final gate decision.
Review the changes since a fixed point (commit, branch, tag, or merge-base) along two axes — Standards (does the code follow this repo's documented coding standards?) and Spec (does the code match what the originating issue/PRD asked for?). Runs both reviews in parallel sub-agents and reports them side by side. Use when the user wants to review a branch, a PR, work-in-progress changes, or asks to "review since X".
Use when judging a finished user-research prior-art artifact before it is accepted — a research vocabulary map, a per-angle search output, an extract container, or the evidence register and report. An acceptance gate, not authoring. Judges a single-sourced bar: a recorded zero is distinguishable from an unreachable source; a claim stays within what its method could measure; recorded facts match the source; transferability carries a weighable reason and is never folded into certainty; numbers are the source's own; findings in one container are genuinely distinct; population and platform describe the study rather than the project; convergence is across INDEPENDENT sources; certainty is never averaged; absence is phrased as a search result. Approves a thin-but-honest result and revises only on a named, unrecorded gap. Emits exactly VERDICT: approve|revise plus actionable findings.
Complete guide to testing Ruby on Rails applications with RSpec. Use this skill when writing unit tests, integration tests, system tests, or when setting up test infrastructure including factories, shared examples, and mocking strategies.
Create a PR review report. Investigate not only bugs and vulnerabilities, but also check the consistency of the exported conversation context, PR summary, and implemented code, and question the implementer's sanity. Use when the user asks to "write a PR review report," "review the code along with the conversation context," or "question the sanity of this PR."
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.