Isolated Linux desktop workspaces for AI agents — a hidden, agent-owned desktop and browser over MCP, so an agent can do GUI and web work without touching your real desktop.
Generate clean SVG diagrams (flowchart, tree, mindmap, architecture, sequence, quadrant, gantt, state machine, ER, timeline, swimlane, bubble chart, radar chart) from a markdown string or a JSON config via fig(). Auto-layout, zero coordinates needed. Works in browser and Node.js.
Build 3D and AR apps on Apple platforms (iOS, macOS, visionOS) with SceneViewSwift — the SwiftUI wrapper around RealityKit. Use whenever the user asks for "3D in SwiftUI", "AR with ARKit in SwiftUI", a model viewer for iOS, or any Apple-platform 3D/AR app where the dependency is the `SceneViewSwift` Swift Package from `github.com/sceneview/sceneview`. For Jetpack Compose / Android use the `sceneview` skill instead; for the browser use `sceneview-web`. Skip for plain ARKit-SDK / SceneKit / Unity / Unreal / RealityKit-without-SceneViewSwift work.
Baut zu einem Song ein Musikvideo, das vollständig aus Code entsteht, ohne Bild- oder Videomodell. Ein JavaScript-Programm berechnet jedes Bild aus Takt, Lautstärke und Gesang, ein Browser rendert es, ffmpeg setzt das Video zusammen. Verwenden, wenn jemand aus einer Audiodatei (mit oder ohne Liedtext) ein Musikvideo, mehrere Fassungen eines Videos oder ein Erklärvideo ohne Musik aus Code erzeugen will.
Design patches and compositions or develop Windsor's FM music engine (@windsor/engine), song documents, mixer and the browser app (@windsor/app). Use for synth behavior, presets, sequencing, harmony, returns, inserts, the console UI and audio verification.
Verify Phonaria user-facing behavior in the running app with agent-browser, using the project feature map and saved evidence. Use for reproducing UI reports or checking a changed feature; keep existing Playwright checks for CI regressions.
QA control plane for the Luke monorepo. Decides what deterministic evidence is sufficient to consider a change correct, across lint, typecheck, Node unit, real-browser component, Postgres integration and Playwright E2E — then assesses, writes or executes that evidence. Reports PASS / FAIL / SKIPPED / QA GAP with residual risk rather than a bare green tick. Use after implementing a feature or fixing a bug, before committing. Modes: /luke-test (assess, read-only) | write | verify. Scoping: default = diff vs merge-base. /luke-test write apps/api | --since <ref>
Use when the game store entitlement parity workflow needs a safe recovery, rollback, replay, or continuity check to produce a recovery-drill record for the platform entitlement test matrix with starting state, test, and observed outcome. Success means the dry-run or approved non-production drill meets the predeclared recovery target and leaves the baseline intact; purchases and restore behavior match the declared contract in sandbox only. Use configured search, fetch, read, browser, test, and write capabilities only when relevant and authorized. Record evidence, limit refinement to three focused passes, and seek approval before external or irreversible actions.
Run the quality gates for an app in this lab repo (typecheck, lint, tests, build) and look at it in a browser. Use before shipping an app, after changing one, or when asked whether an app works.
Automate terminal UI (TUI) apps with agent-tui for testing, inspection, demos, and scripted interactions. Use when automating CLI/TUI flows, regression testing terminal apps, verifying interactive behavior, or extracting structured text from terminal UIs. Also use when asked what agent-tui is, how it works, or to demo it. Do not use for web browsers, GUI apps, or non-terminal interfaces.
Turn any GitHub repo, local folder, PR, or markdown/Obsidian vault into an interactive architecture map with CodeFlow — a zero-build single index.html browser app (React 18 + D3.js from pinned CDNs) that runs 100% client-side with no backend and no data collection. Pick an input (public repo, private repo with a local token, local files, PR URL, markdown vault), choose a visualization mode (folder/layer/churn/blast), and read the analysis — dependency graph, blast radius, code ownership, heuristic security scanner, pattern/anti-pattern detection, A–F health score, activity heatmap, PR impact — then export JSON/Markdown/text/SVG/PDF or wire the self-updating CodeFlow Card SVG onto a README.
Use when user wants anti-bot stealth browsing, website access that blocks normal Playwright, or browser automation that needs to bypass detection (Cloudflare, reCAPTCHA, FingerprintJS, etc.). CloakBrowser is a source-level patched Chromium with 57 fingerprint patches — drop-in Playwright replacement.
Show the human a running web app — a dev server, a site, a prototype — in their agentbox Preview panel, with live reload. Use whenever they ask to see, preview, run, try or open something you built for the browser, or say "put it in my preview", and before telling them a web app is done.
Guided journey from a shipped app that works but feels rough to a product that fits the job, flows without friction, reads clearly, and persuades honestly. Orchestrates nine skills phase by phase - jobs-to-be-done, ux-heuristics, design-everyday-things, refactoring-ui, microinteractions, made-to-stick, influence-psychology, high-perf-browser, steve-jobs-design-review - asking the user questions at every decision point and recording results in the project docs/ folder (CUSTOMER.md, DESIGN.md, POSITIONING.md, IMPROVE-APP-PLAN.md) so the journey resumes across sessions. Use when the user wants to fix a clunky product, cut UX friction, sharpen in-app copy and prompts, or says 'the app works but feels rough'. Do not use for code, tests, or production hardening - use improve-code-quality (fresh prototype) or remove-technical-debt (aged codebase); no app yet, create-app; needs growth loops, grow-app; marketing-site friction, improve-website. For one framework in isolation, invoke that skill directly.