发现 Skills

Create new AI chat interface components for the ai-elements library following established composable patterns, shadcn/ui integration, and Vercel AI SDK conventions. Use when creating new components in packages/elements/src or when the user asks to add a new component to ai-elements.

Run the QAMAS CLI (`qamas ask`) on the user's question and return the exact stdout.

Discovery phase - research and codebase exploration

Regenerate the TypeScript API client from the OpenAPI specification after backend API changes. Run this after modifying backend/src/main/resources/api/openapi.yaml.

Comprehensive shadcn/ui component library with theming, customization patterns, and accessibility. Use when building modern React UIs with Tailwind CSS. IMPORTANT - Always use MCP server tools first when available.

Create or update project-independent technical notes in a personal knowledge base (`~/zettel/`). Records research findings and insights in a form your future self can revisit. Usage: /zettel <content or instruction>

Entry point for design tasks. Routes to brand-guidelines (brand identity, voice, assets), design-system (tokens, component specs), or ui-styling (implementation). Use for design decisions, visual consistency, design system creation.

Complete slide creation toolkit for Marp/Marpit presentations. Use when creating presentations, authoring slides, writing slide content, drawing diagrams, creating illustrations, designing slide color schemes, choosing presentation colors, designing slide themes, selecting background/text/accent colors, or any combination of these tasks. Covers Markdown authoring (Marpit/Marp), SVG illustration, and color design for technical presentations, PowerPoint, Keynote, architecture diagrams, and developer-focused decks.

Integrate the @navikt/lumi-survey feedback widget into a NAV frontend application

Hermes Agent project architecture — the relationships between cli.py, run_agent.py, HermesCLI, and AIAgent, and how to correctly add new commands and handlers. Use when modifying the Hermes codebase, adding new commands, or understanding Hermes internals.

Provides guidance for LLM post-training with RL using slime, a Megatron+SGLang framework. Use when training GLM models, implementing custom data generation workflows, or needing tight Megatron-LM integration for RL scaling.

Guided .NET Framework to .NET 8 migration workflow. Use when: converting .NET Framework 4.x projects to .NET 8, upgrading csproj files, migrating Startup.cs to minimal hosting, converting EF6 to EF Core, updating NuGet packages, modernising ASP.NET MVC to ASP.NET Core.

CLI modernization and hooks system enhancement for claude-flow v3. Implements interactive prompts, command decomposition, enhanced hooks integration, and intelligent workflow automation.

Dependency policy: latest-compatible, source-backed, SLSA Level 2, SBOM, lockfile discipline. Use for: dependencies, versions, package updates, version migration, compatibility, supply-chain. Triggers: bump dependencies, package version policy, latest-compatible, lockfile discipline, supply chain check, SLSA Level 2, SBOM.

Initialize a project's context: read-only scoped discovery with Serena, fullrepo bootstrap, architecture, report. Use for: study the project, initialize context, analyze the repository, scope/sphere/module discovery. EN triggers: ry-init, init project, init scope, initialize context, project onboarding, context pack, study repo, learn project, fullrepo bootstrap.

Audit implementation against the rldyour rules. Use for: /ry-rules-review, check against rules, rules audit, check hard rules, quality-by-rules checks. EN triggers: rules review, hard rules audit, check against rules, rldyour rules check, policy compliance audit, rules audit.

Professional copywriting squad with 13 specialized commands (headline, bullets, sales letter, VSL, email sequence, landing page, ad copy, funnel copy). Consumes RESUMO-SCALE.md and LINGUAGEM-CRUA.md from the avatar to calibrate tone of voice and the client's real phrases. Phase 6 of the Código ZERO flow.

Google Ads campaign orchestrator. Creates, analyzes and optimizes Search, Shopping and Performance Max campaigns with a single command. Pipeline: business analysis, keywords, RSA ads, targeting, negatives, performance. All via direct API (google-ads-api / Opteo), no dashboard, no CSV. Actions: create campaign, analyze search terms, add negatives, generate RSA ads, pause, activate, adjust budget, Thomas report. Topics: Google Ads, campaign, Search, PMax, Shopping, keywords, negatives, search terms, RSA, headline, description, Quality Score, ROAS, CPA, budget, bid strategy.

Control Philips Hue lights and scenes via the OpenHue CLI.

Automatically creates new skills from complex problem-solving sessions. Triggers at the end of multi-step tasks that required novel approaches, debugging sessions, or workflow creation. Use when: a task took 3+ tool calls, involved a non-obvious solution, or established a pattern likely to recur. Also triggers on: "create a skill from this", "save this as a skill", "this should be a skill".

Manage character visual identities. Create characters, build reference libraries from seed images and turnaround variations, and generate consistent images on demand via the NeukoAI Character Image Studio API.

Five Whys root cause analysis. Iteratively asks "why" to drill past symptoms to underlying causes. Use for debugging, investigating failures, or understanding why something went wrong.

MoSCoW prioritization categorizing items as Must have, Should have, Could have, or Won't have. Use for scope definition, feature prioritization, or when everything feels equally important.

Start-Stop-Continue retrospective identifying what to Start doing, Stop doing, and Continue doing. Use for sprint retros, personal reflection, team process reviews, or habit audits.