Windags Skills

Build audio transcription pipelines with Whisper, Deepgram, and AssemblyAI including speaker diarization and real-time streaming. Activate on: transcription, speech-to-text, diarization, audio processing, meeting transcripts. NOT for: text-to-speech synthesis (voice-audio-engineer), music generation (ai-engineer).

Building resilient distributed systems with circuit breakers, retries with full-jitter exponential backoff, retry budgets (per-request 3-attempt + per-client 10% ratio per Google SRE), deadline propagation, and the cascading-failure math (4 layers × 3 retries = 64x amplification). Grounded in Resilience4j, Microsoft Cloud Patterns, AWS Architecture Blog (Marc Brooker), and Google SRE Book.

Use when wrangler deploys silently fail or produce wrong artifacts, secrets upload as empty strings, custom domain DNS is not resolving, route assignment broke after rename, observability/tail logs are needed, D1/KV/R2 bindings are missing, OAuth scope errors block a command, cookies on a redirect are not attaching, or the worker exceeds CPU time. Triggers: "Tenant or user not found", "Authentication error" on wrangler d1 push, secret length is 0 after upload, route returns origin not worker, "compatibility_date too old", 1101 errors, custom_domain assignment lost on script rename. NOT for AWS Lambda / Vercel Edge / Deno Deploy (different runtimes), Cloudflare Zero Trust, or R2-specific tuning.

Generates starter component code using design tokens. Creates React/Vue/Svelte components with proper token usage, variants, and accessibility built in.

Audit designs, color palettes, and UI components for WCAG 2.1 accessibility compliance, providing specific fixes for contrast, touch targets, focus indicators, and screen reader support. NOT for color harmony aesthetics or design system creation.

Multi-stage Docker build optimizer for minimal, secure production images. Activate on: Dockerfile optimization, multi-stage build, distroless image, container size reduction, Docker security scanning, BuildKit features. NOT for: container orchestration (use kubernetes-manifest-generator), CI/CD pipelines (use github-actions-pipeline-builder), runtime container config (use environment-config-manager).

Benchmark multi-agent coordination systems with experiment design, power analysis, human rating protocols, bootstrap confidence intervals, and reproducible reporting. Use for salvage latency, recovery fidelity, coordination overhead, and protocol comparison studies. NOT for ML model benchmarking, web-product A/B testing, survey design, or general-purpose data science.

Structured interview and observation techniques for extracting tacit knowledge from domain experts

Use for sensemaking, frame revision, anomaly handling, questioning strategy, and ambiguity diagnosis with the Data/Frame model before action is chosen. NOT for deterministic classification, settled execution planning, or situations where the frame is already stable.

Use when building server-driven UIs with htmx, designing hx-* attribute compositions, choosing swap strategies (innerHTML/outerHTML/morph), wiring out-of-band updates, integrating with form validation, or migrating from a SPA back to server-rendered HTML. Triggers: hx-get/hx-post setup, hx-target + hx-swap interplay, hx-trigger debouncing, OOB swaps for parts of the page outside the request target, response headers like HX-Trigger and HX-Push-Url, integration with Hyperscript or Alpine.js, accessibility considerations on partial updates. NOT for full SPAs (use React/Vue), framework-specific server libraries (use those skills directly), or non-HTML responses.

Use when triaging CrashLoopBackOff, OOMKilled, ImagePullBackOff, Pending pods, networking failures, ingress 502/504, PVC stuck, HPA not scaling, init-container loops, sidecar startup ordering, or "kubectl describe says nothing useful". Triggers: pod restart counts climbing, "no nodes available to schedule", "back-off restarting failed container", DNS resolution failures inside the cluster, NetworkPolicy denials, livenessProbe killing healthy pods. NOT for cluster setup (kubeadm, kops), cloud-provider-specific managed K8s admin (EKS/GKE/AKS), helm chart authoring, or service mesh internals.

Shutting down HTTP services in Kubernetes without dropping traffic — the SIGTERM/preStop dance, the EndpointSlice removal race, the `preStop sleep` pattern, terminationGracePeriodSeconds budgeting, and the Node.js + Go server skeletons. Grounded in kubernetes.io and language runtime docs.

Mobile biometric authentication expert for Face ID, Touch ID, BiometricPrompt, Keychain/Keystore, and WebAuthn. Activate on: biometric authentication, Face ID, Touch ID, BiometricPrompt, Keychain, Keystore, WebAuthn, passkeys, FIDO2, device authentication. NOT for: OAuth/OIDC flows (use oauth-oidc-implementer), secret management (use secret-management-expert), general security (use security-auditor).

Field-based research methodology studying expert decision-making in real-world high-stakes environments

Use when implementing or reviewing OAuth 2.0 / OAuth 2.1 / OpenID Connect from scratch in a real codebase, choosing a flow (authorization code + PKCE, client credentials, BFF), validating ID tokens, storing tokens safely in browsers, sizing refresh-token rotation, or migrating off implicit / ROPC. Triggers: "should I use a JWT or session cookie", PKCE code_challenge/code_verifier, exact redirect_uri match, state vs nonce confusion, ID-token replay, refresh-token rotation, BFF (backend-for-frontend) pattern, token in localStorage warning. NOT for SAML / WS-Federation, building an authorization server (use a battle-tested IdP), passwordless-only flows (passkeys/WebAuthn), or session-cookie auth without a third-party IdP.

Expert project manager for ADHD engineers managing multiple concurrent projects. Specializes in hyperfocus management, context-switching minimization, and parakeet-style gentle reminders. Activate on 'ADHD project management', 'context switching', 'hyperfocus', 'task prioritization', 'multiple projects', 'productivity for ADHD', 'task chunking', 'deadline management'. NOT for neurotypical project management, rigid waterfall processes, or general productivity advice without ADHD context.

Use when designing or fixing the retrieval side of a RAG system, choosing chunking strategy (fixed-size / recursive / semantic), implementing hybrid search (BM25 + dense) with RRF fusion, adding a cross-encoder reranker, evaluating with RAGAS, or running an index-freshness pipeline. Triggers: "RAG keeps citing the wrong doc", chunk size 512 tokens with overlap, RRF reciprocal rank fusion, dense+sparse hybrid, cross-encoder rerank top-5, RAGAS faithfulness / context precision, voyage-3-large vs text-embedding-3-large, daily / hourly reindex cadence. NOT for fine-tuning vs RAG decision (separate skill), agentic tool-use designs, vector DB operational tuning, or general LLM prompt engineering.

AST-based semantic conflict prediction for multi-agent coding environments. Uses tree-sitter to parse code into ASTs, extract symbol-level claims, build dependency graphs, and predict conflicts BEFORE they happen -- at work-assignment time, not merge time. The key insight: git's textual merge is necessary but not sufficient. Two changes can merge cleanly and break the program. This skill covers tree-sitter fundamentals, symbol-level claims, dependency graph construction, conflict prediction algorithms, and integration with Port Daddy or similar coordination daemons. NOT FOR: textual merge conflict resolution (use git skills), linting or formatting (use static analysis tools), runtime verification of invariants (use runtime-verification-for-agents), or general multi-agent orchestration (use multi-agent-coordination).

ShadowBox training method for accelerating expertise transfer through expert decision comparison exercises

Kafka Streams, Flink, Spark Streaming, and CDC for real-time data pipelines. Activate on: streaming, Kafka Streams, Flink, Spark Streaming, CDC, Debezium, real-time pipeline, event stream processing. NOT for: message broker setup (use event-driven-architecture-expert), batch processing (use batch-processing-optimizer).

Big tech ML engineer to indie founder transition coach. Expert in idea validation, MVP development, marketing, monetization, and sustainable growth for ADHD entrepreneurs. Activate on 'entrepreneur', 'indie founder', 'startup', 'MVP', 'monetization', 'big tech to indie', 'ADHD business', 'app launch', 'side project'. NOT for neurotypical entrepreneurship, VC-backed startups, or traditional business consulting without ADHD context.

Advanced TypeScript type system patterns for production codebases. [What: branded types for nominal typing, discriminated unions, template literal types, conditional types, the infer keyword, satisfies operator, const assertions, Zod schema inference, type-safe event emitters, exhaustive switch checking] [When: designing domain models, building type-safe APIs, creating reusable generic utilities, eliminating runtime bugs with compile-time guarantees, refactoring any-typed codebases] [Keywords: branded types, discriminated union, template literal types, conditional types, infer, satisfies, const assertion, Zod inference, exhaustive, mapped types, utility types, nominal typing, type narrowing, generic constraints] NOT for basic TypeScript syntax or React component typing (use a React-specific skill).

Summarizes very long texts (books, handbooks, biographies, codebases) using hierarchical multi-pass extraction with cheap model armies. Produces structured knowledge maps, not just summaries. Use when processing 50+ page documents, professional handbooks, career biographies, or any text too large for a single context window. Activate on "summarize book", "summarize handbook", "long document", "extract knowledge", "distill text", "professional biography". NOT for short text summarization (<10 pages), real-time chat summarization, or code documentation (use technical-writer).

Build background AI agents that run alongside developers during vibe coding sessions, proactively helping without being asked. Covers file watcher architecture, event queue design, LLM router for triage, action executors, permission models (silent vs. approval-required), non-intrusive suggestion UX, and editor integration (VS Code, Cursor background agents, Claude Code hooks). Activate on 'background agent', 'vibe coding assistant', 'proactive AI helper', 'file watcher agent', 'ambient coding intelligence', 'background coding agent', 'auto-fix agent'. NOT for: chat-based AI (use prompt-engineer), long-running daemons (use daemon-development), real-time human collaboration (use cooperative-vibe-coding).

AI-powered calendar management and agent-based scheduling coordination. Covers calendar APIs (Google Calendar, CalDAV/iCal), AI scheduling assistants (Reclaim, Clockwise, Motion, Cal.com), building custom calendar agents with MCP, multi-calendar merging, timezone management, focus block protection, meeting fatigue detection, and agent-to-agent meeting negotiation protocols. Activate on: "calendar agent", "AI scheduling", "calendar coordination", "meeting scheduling", "calendar API", "focus time protection", "calendar optimization", "Google Calendar MCP", "Reclaim", "Clockwise", "Motion", "Cal.com", "smart scheduling", "calendar-aware agent", "timezone scheduling", "agent negotiation meetings". NOT for: manual calendar UI component design (use form-validation-architect), project management scheduling or Gantt charts (use project-management-guru-adhd), general time-tracking or pomodoro apps (use adhd-daily-planner for time-awareness), building the agent itself from scratch (use agent-creator).

Prepare for Anthropic-specific technical interviews covering Constitutional AI, RLHF, interpretability, scaling laws, MCP, agentic systems, and AI safety. Activate on "anthropic interview", "constitutional AI prep", "alignment interview", "safety interview", "RLHF opinion", "interpretability prep", "anthropic technical". NOT for general ML interview prep, system design interviews, coding challenges, or behavioral interview practice.

Identify and counteract cognitive biases in medical decision-making through systematic error analysis and contextual algorithm application. For diagnostic reasoning, treatment decisions, and clinical judgment improvement. NOT for basic medical knowledge, technical procedures, or non-clinical healthcare domains.

CQRS pattern, event stores, projections, eventual consistency. Activate on: CQRS, event sourcing, event store, read model, projection, aggregate, domain event, command handler. NOT for: message broker setup (use event-driven-architecture-expert), database optimization (use data-warehouse-optimizer).

Generalized Partial Global Planning with TAEMS task structures for multi-agent coordination

Diagnose and eliminate the GraphQL N+1 resolver explosion using Facebook DataLoader's per-request batching, plus the second-order defenses (persisted queries, depth limits, cost analysis). Use when a GraphQL query that returns N items in a list triggers N+1 database round trips, when designing nested resolvers that load associations, when a list query times out under load, or when you need to budget query cost before resolvers run. NOT for REST-style endpoint design (use api-versioning-strategy), generic database query optimization (use database-query-optimizer), or schema design (use graphql-server-architect).

Monetization strategies for indie developers, solopreneurs, and small teams. Covers freemium models, SaaS pricing, sponsorships, donations, email list building, and passive income for developer tools, content sites, and educational apps. Activate on 'monetization', 'make money', 'pricing', 'freemium', 'SaaS', 'sponsorship', 'donations', 'passive income', 'indie hacker'. NOT for enterprise sales, B2B outbound, VC fundraising, or large-scale advertising (use enterprise/marketing skills).

GOMS cognitive modeling methodology for analyzing human-computer interaction task performance

Comprehensive survey of knowledge distillation methods, architectures, and applications in neural network compression

Kubernetes manifest and Helm chart generator for production workloads. Activate on: K8s config, Deployment YAML, HPA autoscaling, PodDisruptionBudget, Ingress rules, NetworkPolicy, Helm chart, Kustomize. NOT for: Docker image building (use docker-multi-stage-optimizer), IaC provisioning of clusters (use terraform-module-builder), CI/CD pipeline config (use github-actions-pipeline-builder).

Master of high-performance web map implementations handling 5,000-100,000+ geographic data points. Specializes in Leaflet.js optimization, Supercluster algorithms, viewport-based loading, canvas rendering, and progressive disclosure UX patterns.

Mobile offline-first architecture with local databases, CRDT conflict resolution, and background sync. Activate on: offline sync, offline-first, local database, WatermelonDB, SQLite, CRDT, conflict resolution, background sync, mobile persistence. NOT for: server-side databases (use data-pipeline-engineer), web caching strategies (use pwa-architect), API design (use api-architect).

Use when adding distributed tracing, debugging missing spans, fixing W3C traceparent propagation, configuring OTLP exporters (gRPC vs HTTP), choosing sampling strategies, setting resource attributes, or wiring auto-instrumentation libraries. Triggers: spans missing in Datadog/Honeycomb/Jaeger, "service.name = unknown_service", trace assembly broken across services, async work losing context, OTLP collector unreachable, sampling rate decisions, ESM vs CJS auto-instrumentation loader bugs. NOT for vendor-specific SDKs (Datadog APM, New Relic), structured-logging-only setups, or pre-OTel tracers (Jaeger client, Zipkin Brave).

Use when you need to publish a message to Kafka / RabbitMQ / SNS atomically with a database commit, when fixing the dual-write hazard between DB and broker, deciding between polling-publisher and CDC (Debezium) relay, designing the outbox schema, handling at-least-once / idempotency on the consumer, or pruning a high-volume outbox table. Triggers: dual write, "we updated the DB but the event never published", outbox/inbox pattern, Debezium connector, EventRouter SMT, aggregate_type / aggregate_id, at-least-once with idempotent consumer, outbox table partitioning. NOT for receiver-side webhook handling, sagas/Temporal as orchestration, log-based event sourcing as the source of truth, or Kafka producer tuning generally.

Generate pixel art diagrams and infographics for recovery education articles in retro 16-bit game aesthetic

Implement paywalls, subscription billing, and feature gating in web applications. Covers Stripe/Lemon Squeezy/Paddle integration, subscription lifecycle management, React/Next.js gating patterns, webhook handling, and pricing page design. Activate on 'paywall', 'subscription', 'billing', 'Stripe integration', 'feature gating', 'pricing tier', 'payment', 'monetize', 'charge users', 'Stripe Checkout', 'webhook handler', 'subscription state'. NOT for pricing strategy decisions (use indie-monetization-strategist), not for payment UI design only (use web-design-expert), not for in-app mobile purchases (App Store/Play Store rules are a different beast).

App Store and Google Play submission automator with Fastlane, screenshot automation, metadata management, and TestFlight/internal testing. Activate on: app store submission, Fastlane, TestFlight, Google Play Console, screenshot automation, metadata management, app review, code signing, provisioning profiles. NOT for: CI/CD pipeline setup (use github-actions-pipeline-builder), app architecture (use react-native-architect), analytics (use mobile-analytics-crash-reporting-expert).

Use when optimizing Docker builds with BuildKit, designing multi-stage Dockerfiles, leveraging cache mounts, building multi-arch images, choosing distroless vs alpine, or signing images with cosign. Triggers: docker build slow, layer cache not hitting, npm install / pip install reruns, multi-stage final image bloat, glibc vs musl issues, RUN --mount=type=cache, BUILDKIT_INLINE_CACHE, registry cache exporters, qemu emulation for arm64, COPY ordering for cache. NOT for Docker daemon admin, Kubernetes-specific image policies, container security scanning workflows, or Buildpacks (different paradigm).

Use when designing matrix builds (OS x version x dimension), composing reusable workflows, gating jobs on conditions, doing OIDC to AWS/GCP/Cloudflare without long-lived secrets, sharding tests across runners, or fixing cache invalidation. Triggers: workflow_call between repos, fail-fast: false tradeoffs, exclude/include matrix surgery, environment protection rules, OIDC trust policy setup, concurrency cancellation on duplicate pushes, artifact reuse across jobs, conditional job matrices computed at runtime. NOT for GitLab CI / CircleCI / Jenkins (different yaml dialects), self-hosted runner administration, or local act dev.

Generate hand-drawn style diagrams and infographics for recovery education articles. Creates anatomist's notebook aesthetic visuals - brain diagrams, timelines, social comparisons, and process flows using continuous line art, semantic color coding, and margin annotations.

Build production LLM streaming UIs with Server-Sent Events, real-time token display, cancellation, error recovery. Handles OpenAI/Anthropic/Claude streaming APIs. Use for chatbots, AI assistants, real-time text generation. Activate on "LLM streaming", "SSE", "token stream", "chat UI", "real-time AI". NOT for batch processing, non-streaming APIs, or WebSocket bidirectional chat.

Mobile deep linking specialist for Universal Links, App Links, deferred deep links, and attribution. Activate on: deep linking, Universal Links, App Links, deferred deep link, app attribution, URL scheme, branch.io, dynamic links, app clip. NOT for: web routing (use frontend-architect), push notification handling (use mobile-push-notification-expert), API URL design (use api-architect).

Generate cross-modal embeddings with CLIP, SigLIP, and ImageBind for text-image-audio search. Activate on: multimodal search, text-to-image search, cross-modal embeddings, CLIP embeddings, visual search. NOT for: text-only embeddings (ai-engineer), image classification (computer-vision-pipeline).

Build offline-first web apps with Service Workers, Workbox, background sync, and progressive enhancement strategies. Activate on: offline support, service worker caching, background sync, app shell, cache strategies, precaching. NOT for: native app packaging (use tauri-expert), general PWA manifest/install (use pwa-expert).

Next.js App Router RSC architecture, Server Actions, streaming SSR, and partial hydration. Activate on: 'use server', 'use client', server components, App Router, streaming, partial hydration, Server Actions. NOT for: Pages Router (use nextjs-pages-router), client-only SPAs (use react-performance-optimizer), API routes without UI (use api-architect).

Expert in building blazing-fast reactive dashboards with comprehensive testing. Masters React performance patterns, testing strategies for async components, and real-world patterns from Linear, Vercel, Notion. Specializes in sub-100ms dashboard loads, proper test mocking, skeleton states, and optimistic updates.

Build real-time collaborative editing with WebSockets, OT/CRDT conflict resolution, and presence awareness. Implements cursor tracking, optimistic updates, and offline sync. Use for collaborative editors, whiteboards, video editing. Activate on "real-time collaboration", "WebSocket sync", "multiplayer editing", "CRDT", "presence awareness". NOT for simple chat, request-response APIs, or single-user apps.

Long-lived transaction pattern using compensation-based rollback for distributed system consistency

Use when integrating Hugging Face Transformers.js (Xenova/transformers) for in-browser or in-Node inference, debugging quantized model loading, building bi-encoder / cross-encoder / classification pipelines, configuring model cache directories, or bypassing high-level pipelines to read raw logits. Triggers: "model failed to load", cross-encoder scores all 1.0 (softmax-over-1 trap), env.allowLocalModels confusion, cacheDir overrides, ONNX runtime mismatch, ESM vs CJS pipeline imports, browser vs node feature gaps. NOT for full transformers Python (different SDK), TensorFlow.js, ONNX Runtime Web directly without Transformers.js, or model training.

Migrate vector data between Pinecone, Qdrant, Weaviate, pgvector with re-embedding and schema mapping. Activate on: vector DB migration, switch vector database, re-embed collection, migrate embeddings. NOT for: initial ingestion (rag-document-ingestion-pipeline), embedding model training (ai-engineer).

Use for insight, reframing, contradiction, impasse, and anomaly-driven problem solving when execution effort no longer helps. NOT for routine optimization, error correction, or well-specified tasks with known solution paths.

Implement WCAG 2.2 AA/AAA compliance with automated testing, keyboard navigation, screen reader support, and focus management. Activate on: accessibility audit, WCAG compliance, keyboard navigation, screen reader, aria attributes, axe-core, focus trap. NOT for: design-level accessibility review (use design-accessibility-auditor), color contrast only (use css-in-js-architect).

Use when designing or fixing a Content Security Policy on a real site, choosing between nonce-based and hash-based CSP, adding strict-dynamic, debugging "Refused to execute inline script" errors, deploying CSP in report-only mode first, configuring report-to / report-uri, or auditing an existing policy for unsafe-inline / unsafe-eval / wildcards. Triggers: "CSP blocks legitimate inline script", strict-dynamic, nonce-{RANDOM}, sha256-{HASH}, object-src none, base-uri none, frame-ancestors, Trusted Types, X-Content-Security-Policy obsolete, report-only vs enforced. NOT for general HTTP security headers (HSTS, COOP/COEP), Trusted Types deep dive, CORS configuration, or building a WAF.

Choose, compare, and decompose work into the right coordination topology: dag, workflow, manager-driven team, swarm, blackboard, team-builder, or recurring loop. Use when the coordination shape is unclear, when you need topology-native breakdown instructions instead of a generic DAG, or when planning topology and runtime topology may differ. NOT for actual execution, node-level skill matching, or framework/tool installation.

Design desktop window systems that feel native on macOS and Windows: split views, inspectors, auxiliary windows, tiling, snap-safe sizes, default placement, and chrome density. Use when a Tauri/Electron/WinUI/SwiftUI app feels awkward, overlapping, over-chromed, or generic, or when you need a real geometry contract for multiple surfaces. NOT for mobile layouts, marketing pages, or CSS-only component styling without multi-window or multi-pane behavior.

Implementing distributed tracing with W3C Trace Context — the byte-precise `traceparent` and `tracestate` header formats, OpenTelemetry's W3CTraceContextPropagator as the modern default (replacing X-B3-* and X-Datadog-*), head vs tail sampling, and HTTP→SQL trace propagation via sqlcommenter. Grounded in the W3C REC and OpenTelemetry specs.