<!-- The flow map below is derived from mattpocock/skills' ask-matt (MIT, see THIRD_PARTY_NOTICES.md). The route is AOS's own: most stations on upstream's main flow have no AOS equivalent, so copying it verbatim would have produced a router pointing at skills that do not exist. -->

ask-tim: which way do I go?

You don't remember every skill, so ask. This does no work itself.

Two layers, answering different questions. The flow map answers "which way?" — the route work travels from an idea to something shipped, and which branch to take at each fork. The catalogue below it answers "which skill?" — grouped by domain, with overlap guidance for the cases where several look alike. Start with the flow; drop to the catalogue when you already know roughly where you are.

---

The main flow: idea → ship

The route most work travels.

1. Sharpen the idea. /grill-with-docs whenever you are in a working directory — the interview plus a paper trail, writing settled terminology into CONTEXT.md and ADRs as it goes. Without a repo to write to, /grill-me. Both run the same grilling primitive; the difference is only whether the understanding survives the session.

For a bigger or fuzzier idea, /bdbrainstorm runs the same interview inside a multi-agent debate (/bdbmediastorm for live show-control and event tech). Don't grill twice — those skills already do this step.

2. Branch: how much machinery does the build need? This is the fork that matters, and picking wrong is expensive in both directions.

• /startcycle — the linear chain, and the right answer for most daily work. Architect → TechLead → Build → Reviewer, hand-offs as plain files, no state machine.
• /startcycle-graph — when you need the durable state.json, a Reviewer repair loop with a no-progress guard, an automated quality gate, and escalation to a human. Mission-critical or multi-session work. Runs headless.
• /startcycle-graph-user — a throwaway 2–4 node fan-out for one-off parallel work. Nothing persistent left behind.
• Two-file edit? No pipeline. Just do it.

3. Review the plan before it is built. At the Architect→TechLead gate, aos-plan-canvas open production_artifacts/00_execution_plan.md opens the plan in a browser where you point at what should change instead of retyping it. Mandatory at the end of /bdbrainstorm and /bdbmediastorm; optional in /startcycle; deliberately not forced in /startcycle-graph, which runs headless.

4. Build, then review adversarially. The pipelines invoke their own build and review nodes — you do not call these by hand. Reviewer's pass is a correctness check against the plan's contract; godmode-shipping's gate is mechanical (lint, typecheck, tests). Two different checks, deliberately not merged.

Reach for test-driven-development on their own when you want one behaviour built test-first without a whole pipeline, and git-pr-review when reviewing a branch or PR against a fixed point.

Forcing a specific skill into a run: --skill=<name> on any startcycle variant makes it a hard requirement — for a private skill of your own no node would otherwise reach for. Validated before the run starts; a name that does not resolve halts rather than proceeding without it.

Outside the core: the ECC store

When no native AOS skill fits, search the markdown-only ECC store before starting a pipeline:

• aos-store search <query> finds available fallback skills and agents.
• aos-store install <name> --net installs a verified item after explicit user action.
• aos-store list --type=skills and aos-store list --type=agents show the catalogue.
• The core remains the trusted AOS skill set; the store is a fallback, not a reason to bypass the startcycle validation.
• A missing --skill is reported by the dispatcher with the exact install command. It never downloads during a run.

---

On-ramps

A starting situation that generates work, then merges onto the main flow.

• Issues piling up → /triage. For work you did not create: bug reports, incoming requests, anything raw. Tickets a pipeline already produced are agent-ready — do not triage them.
• Something is broken → systematic-debugging for a structured hunt, debugger for a plain error. Both refuse to theorise before there is a reproduction.
• You need a runnable answer, not an argument → /prototype. Throwaway code to settle a question that conversation cannot, then bring what you learned back to the idea thread.
• Deploying or shipping infrastructure → godmode-shipping gates a release; bdb-deploy, vercel-deployment and cloudflare-workers-expert do the actual shipping.

---

What this flow does not cover

Named so you don't go looking: AOS has no equivalent of a cross-session handoff file, and no wayfinder-style skill for charting a months-long effort as decision tickets. For work too large for one session, /startcycle-graph's durable state.json is the closest thing — it is resumable, but it does not chart the fog for you.

---

🧭 Intent Index (Table of Contents)

If the user's intent matches one of these, jump to the corresponding section:

• "I need to choose a build pipeline (startcycle vs startcycle-graph)" ➔ Agent Orchestration & Pipelines
• "I need multi-agent brainstorming or ideation" ➔ Agent Orchestration & Pipelines
• "I want to build a new UI component or page" ➔ Frontend, UI & Motion
• "I want Tailwind, shadcn, or design-system UI work" ➔ Frontend, UI & Motion
• "I'm optimizing React or Next.js performance" ➔ Frontend, UI & Motion
• "I'm designing an API (REST/GraphQL)" ➔ Backend, APIs & Databases
• "I need Prisma, Drizzle, or Postgres schema help" ➔ Backend, APIs & Databases
• "I'm working with n8n workflows" ➔ Backend, APIs & Databases
• "I want to deploy to Vercel, Cloudflare, or a VPS" ➔ DevOps, Git & Deployment
• "I'm managing GitHub PRs, issues, or monorepos" ➔ DevOps, Git & Deployment
• "I want to scrape or extract data from a website" ➔ Web Scraping & Browser Automation
• "I need to run browser automation or web testing" ➔ Web Scraping & Browser Automation
• "I'm building an AI agent, RAG pipeline, or working with LLMs" ➔ AI, LLMs & Agents
• "I need an SEO, GEO, or accessibility audit" ➔ Content, SEO & Docs
• "I need to write a README or documentation" ➔ Content, SEO & Docs
• "I need to debug a failing test or weird bug" ➔ Core Programming & Debugging
• "I want to ship a feature safely to production" ➔ Core Programming & Debugging
• "I'm managing the BDB SaaS multi-cloud fleet" ➔ BDB Ecosystem & SaaS Ops
• "I need Three.js, 3D, or motion work" ➔ Media & EventTech
• "I need live event tech, TouchDesigner, or Resolume help" ➔ Media & EventTech
• "I need PCB layout, schematic capture, or electrical/hardware design" ➔ Electrical & Hardware Design

---

🏗️ Agent Orchestration & Pipelines

Use these skills to orchestrate complex multi-agent workflows.

• Top Picks: startcycle-graph, bdbrainstorm

Overlap: The 3 Build Pipelines

Don't guess which pipeline to use. Follow these rules (source of truth: CLAUDE.md and skills/basic/*/SKILL.md):

• startcycle: Use for a straight-line, predictable run through the agents with file hand-offs and no state machine. Cheapest of the three.
• startcycle-graph: Use when you need the full dispatcher graph: durable state.json, a Reviewer repair loop, a quality gate, and human escalation.
• startcycle-graph-user: Use for a small throwaway fan-out (2-4 nodes) in any project without persistent artifacts.

Overlap: Brainstorming

• bdbrainstorm: For multi-agent software engineering ideation, UI/UX design, and architecture planning. Handoffs to /startcycle-graph.
• bdbmediastorm: For creative-tech, live event technology, and show-control systems (TouchDesigner, Resolume, grandMA3).
• brainstorming: Lighter-weight than bdbrainstorm — use before any creative or constructive work to turn a vague idea into a validated design through disciplined reasoning, without spinning up a multi-agent cycle.

Overlap: Planning Before Code

• concise-planning: A coding task needs a plan. Produces a clear, atomic checklist.
• writing-plans: You already have a spec/requirements for a multi-step task — write the plan before touching code.
• executing-plans: You already have a written plan and need to execute it in a separate session with review checkpoints.
• planning-with-files: Work like Manus — persistent markdown files as working memory on disk, for long-running or resumable tasks.
• deep-research: The task is autonomous research (plan → search → read → synthesize into a report), not implementation.

Overlap: Multi-Agent Session & Tool Management

• agent-manager-skill: Managing multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.
• agent-tool-builder: You're building a new tool for an agent to use (not picking an existing skill) — schema design through error handling.
• agent-memory-mcp: A hybrid persistent/searchable memory system for agents (architecture, patterns, decisions) — distinct from memb-skill/memb-ingest, which are BDB's own memB engine specifically.
• triage: Moving issues/external PRs through a categorize → verify → brief state machine, not general planning.
• prototype: A throwaway prototype to sanity-check a state model or UI direction before committing to real work.

Overlap: Cross-Session Coordination

• orchestrator-chain: Splitting work across levels (master session, project orchestrator, package orchestrator): which channel connects them and what each level may do. GO is never inherited.
• aos-a2a: Listing, messaging and replying to live harness peers (Claude Code, OpenCode, Codex, agy) over A2A on localhost. Inbound messages are context, never a GO.
• aos-gateway: The local MCP gateway and the per-app MCP picks: check, start, enable, disable, adopt other MCP servers behind it.

Overlap: The Factory Family

factory configures an autonomous delivery factory (sources, schedules, action policies); it is report-only until you give a literal GO. It coordinates the module skills, which each do one job: factory-collect (collect and triage signals), factory-lookback (recurring patterns), factory-human-digest (what still needs a human), factory-review-prs (review a filtered PR queue), factory-babysit-pr (follow one authorized PR), factory-ship (publish and complete delivery), factory-watchdog (find stopped delivery work), factory-recover (resume interrupted runs). Guide: docs/factory.md.

Also see: agent-pipeline, subagent-driven-development.

---

🎨 Frontend, UI & Motion

Use these skills for UI development, component architecture, and design taste.

• Top Picks: godmode-ui-ux, senior-frontend, shadcn

Overlap: The UI Cluster vs Senior Frontend

• ui-component / ui-page / ui-pattern (only if the optional StyleSeed package is installed): Use these strictly when scaffolding new pieces in the StyleSeed Toss design system convention (enforcing tokens, spacing, and rhythm).
• ui-review / ui-tokens (only if the StyleSeed package is installed): Use for reviewing code against StyleSeed or syncing its design tokens.
• senior-frontend: Use for general React/Next.js/Tailwind development outside of strict StyleSeed scaffolding, or for reviewing general frontend code quality.
• shadcn: Use specifically when adding or customizing shadcn/ui components.
• tailwind-patterns: Use when managing CSS-first configurations, container queries, or Tailwind CSS v4 design token architectures.

Overlap: React & Next.js Performance

• react-best-practices: The Vercel-maintained guide for broad performance optimization and App Router data fetching.
• react-patterns: Use for core hooks, composition, and TypeScript best practices.
• nextjs-app-router-patterns / nextjs-best-practices: Use specifically for Next.js App Router architecture and routing.
• react-component-performance: Use for targeted diagnosis of slow React components (e.g., render bottlenecks).
• tanstack-query-expert: Async state/data-fetching specifically — TanStack Query mutations, stale time, optimistic updates.
• zustand-store-ts: Client-side state stores with Zustand, typed and with middleware.
• web-performance-optimization: Broader than component-level — Core Web Vitals, bundle size, caching across the whole app.
• wcag-audit-patterns: Accessibility compliance auditing against WCAG 2.2, not general UX review.
• frontend-dev-guidelines: General-purpose frontend engineering standards when the task doesn't need a narrower skill above. For the backend half, godmode-engineering and software-architecture.
• landing-page-generator: Specifically a high-converting marketing landing page (PAS/AIDA/BAB copy frameworks), not a general app page.

Overlap: The UX Cluster

• ux-flow (if the StyleSeed package is installed): Designing user flows, progressive disclosure, hub-and-spoke navigation (StyleSeed patterns).
• ux-audit (if the StyleSeed package is installed): Auditing existing screens against Nielsen's heuristics and mobile UX best practices.
• ux-feedback (if the StyleSeed package is installed): Adding loading/empty/error/success feedback states to components.
• ux-persuasion-engineer: Conversion/onboarding/checkout flows specifically — choice architecture, friction audits, commitment design.
• ui-ux-pro-max: Broader design guide (color, typography, general UI review) when the task doesn't fit one of the narrower UX skills above.

Also see: bdbdesignpro, design-spells.

---

🗄️ Backend, APIs & Databases

Use these for server-side architecture, APIs, workflows, and database schema design.

• Top Picks: godmode-engineering, api-design-principles, drizzle-orm-expert

Overlap: Database & ORM Choices

• database-design: Use for overarching schema design, indexing strategy, and general database principles.
• postgres-best-practices: Use for raw performance optimization, queries, and Supabase-specific patterns.
• postgresql: Use for raw schema design, indexing, and Postgres constraints.
• using-neon / neon-postgres: Use specifically when working with Neon's serverless Postgres (branching, connection pooling).
• prisma-expert: Use for Prisma ORM schema design, migrations, and relational modeling.
• drizzle-orm-expert: Use for Drizzle ORM type-safe queries and serverless integrations.

Overlap: The n8n Cluster

• n8n-workflow-patterns: Use for architectural patterns of whole workflows.
• n8n-code-javascript / n8n-code-python: Use when writing custom code inside n8n Code nodes (handles $input/$json nuances for JS/Python).
• n8n-expression-syntax: Use for debugging {{}} expressions.
• n8n-mcp-tools-expert: Use for integrating n8n-mcp tools effectively.

Overlap: Architecture Level

• software-architecture: General quality-focused architecture guidance — the default when the task is "design/analyze architecture" without a more specific fit below.
• microservices-patterns: Specifically service boundaries, inter-service communication, and distributed-systems resilience.
• architect-review: Reviewing an existing architecture against modern practice, not designing a new one from scratch.

Also see: api-patterns, openapi-spec-generation, golang-pro, python-pro.

---

🕷️ Web Scraping & Browser Automation

Use these skills to extract data or control browsers.

• Top Picks: apify-ultimate-scraper, playwright-skill

Overlap: Scraping and Extraction

• apify-ultimate-scraper: The top-level choice. Automatically selects from 55+ Apify Actors for the task.
• apify-lead-generation: Narrow use case: scraping leads from multiple platforms.
• web-scraper: Multi-strategy data extraction (tables, prices) with pagination and CSV/JSON export.
• browser-automation: General principles for selectors, waiting, and anti-detection.
• playwright-skill: Universal executor for custom Playwright code (e2e testing, interactive scraping).
• go-playwright: Use when writing browser automation in Go.
• webapp-testing: Use when writing native Python Playwright scripts for local testing.

---

🚀 DevOps, Git & Deployment

Use these to ship code, manage infrastructure, and handle source control.

• Top Picks: godmode-shipping, github, vercel-deployment

• Godmode Shipping: godmode-shipping is the ultimate pre-launch gatekeeper (lint, tests, rollbacks).

• Deployments: vercel-deployment (Next.js/React), cloudflare-workers-expert (Edge/KV/D1), bdb-deploy (rsync over SSH to VPS), docker-expert (Containers).

• Git/GitHub: git-advanced-workflows, git-pr-review, github-workflow-automation, github-repo, github-actions-templates (CI/CD workflow patterns specifically).

• Monorepos: turborepo-caching, monorepo-management.

---

🧠 AI, LLMs & Agents

Use these when building AI products or optimizing models.

• Top Picks: prompt-engineer, llm-structured-output, rag-engineer

• Agent Dev: ai-agent-development, crewai.

• LLM Integration: gemini-api-integration, gemini-api-dev (Gemini API model capabilities specifically), local-llm-expert, llm-app-patterns.

• Conversational AI Products: llm-application-dev-ai-assistant — building a chatbot/assistant product end-to-end, not just integrating one API.

• Prompting: prompt-engineering-patterns, llm-prompt-optimizer.

• RAG: rag-implementation, rag-engineer (broader — embedding models, vector DB selection), vector-database-engineer (specifically Pinecone/Weaviate/Qdrant/Milvus/pgvector selection and tuning).

• Vercel AI SDK: vercel-ai-sdk-expert — generateText/streamText, UI hooks, tool calling specifically on Vercel's SDK.

---

📝 Content, SEO & Documentation

Use these for docs, markdown, and search engine optimization.

• Top Picks: readme, seo, documentation

• Docs: openwiki-skill, readme, copywriting.

• SEO: seo (broad audit), seo-technical (crawlability, CWV), seo-audit (rankings), geo-fundamentals (AI search), programmatic-seo, schema-markup.

---

🔧 Core Programming & Debugging

Use these for raw problem solving and code hygiene.

• Top Picks: systematic-debugging, clean-code, test-driven-development

• Debugging: debugger (general errors), systematic-debugging (structured triage before fixing).

• Hygiene: simplify-code (diff review for clarity).

• TDD: test-driven-development.

Overlap: Language Expertise

• typescript-pro: Advanced types, generics, strict type safety, enterprise patterns.
• modern-javascript-patterns: ES6+ features and functional patterns (no TypeScript-specific typing concerns).
• python-patterns: Framework selection, async patterns, project structure — decision-making, not a specific bug.
• python-performance-optimization: An existing Python app is slow — profiling and bottleneck fixes specifically.
• go-concurrency-patterns: Goroutines, channels, sync primitives, race-condition debugging.
• bash-linux: General Bash/Linux scripting patterns, macOS or Linux.
• posix-shell-pro: The script must run on any POSIX shell (dash/ash/sh), not just bash.
• tmux: Terminal multiplexing and persistent remote session management specifically.

---

🌀 BDB Ecosystem & SaaS Ops

Specific utilities for the BDB environment.

• Top Pick: memb-skill

Also see: memb-ingest, synapse-integration-skill.

---

🧊 Media & EventTech

Use these for 3D, motion, video and live show control.

• Top Picks: godmode-media-creation (video and montage), godmode-eventtech (live shows), godmode-3d-creation (meshes and scenes)

• Godmodes: Determine the overarching flow (3D, Media, EventTech).

• Implementations: mcp-manage drives the creative applications over MCP, spline-3d-integration (web 3D), threejs-skills (WebGL), remotion (React → MP4, deterministic frame counts).

• Per-application MCP guides: once mcp-manage has told you which server to use, these document the tool surface of one application each — bdb-touchdesigner-mcp, bdb-resolume-mcp, bdb-grandma3-mcp, bdb-davinci-mcp, bdb-adobe-suite-mcp, bdb-after-effects-mcp, bdb-blender-mcp, bdb-unreal-mcp, bdb-rhino-mcp, bdb-vectorworks-mcp. Reach for one only when you already know the application; mcp-manage is the way in.

Which video route?

For video, motion, social posts or scripts, start with the Creator Extension content-studio skill (playbook pb-content-studio): its route(brief) picks Hyperframes or Remotion. The table below is for the cases it does not cover.

The choice is driven by where the pixels come from, not by the output format — all four end in a video file.

Source material	Route
Code-generated motion graphics, text, brand animation	remotion — React components rendered to MP4, deterministic and exactly frame-accurate
Existing footage: cutting, colour, beat-sync	godmode-media-creation + mcp-manage → DaVinci Resolve (davinci-resolve-mcp) or Adobe Premiere (adobe_uxp_mcp)
Compositing, motion design over footage	mcp-manage → After Effects (ae-mcp, after-effects-mcp), or Photoshop for stills (adobe_uxp_mcp)
Generative visuals: audio-reactive, shaders, real-time	godmode-eventtech → TouchDesigner MCP, then record_movie
Generative assets: text→3D, image→3D, AI video	The Creator Extension engines — TRELLIS and TripoSR (3D), Text-to-CAD, OpenMontage (multimodal montage), Video-Shotcraft (shot direction), Palmier-Pro (timeline and grading). The 3D engines run as local CLI scripts (see godmode-3d-creation); comfyui-mcp is only the fallback for TRELLIS without CUDA. Installed as a separate power-up module, not a skill in this catalogue.
Concept not settled yet	bdbmediastorm first — it runs the grilling interview for show-control and media work

Resolve and Premiere are interchangeable at this level: pick whichever is actually installed. Both have a working MCP under mcps/.

---

🔩 Electrical & Hardware Design

Use these for schematic capture, PCB layout, and physical hardware design (KiCad, OpenSCAD).

• Top Picks: godmode-hardware-pcb, code-first-hardware-design, pcb-constraint-definition

• Architectural authority: godmode-hardware-pcb — enforces IPC-standard trace/impedance/stackup math and the headless ERC/DRC/DFM sign-off gate across the skills below; load it before starting schematic or layout work.

• Circuit synthesis & enclosures: code-first-hardware-design — programmatic schematic capture (SKiDL, netlists, S-expressions) and parametric 3D CAD enclosure co-design (OpenSCAD/BOSL2).

• Constraints & stackup: pcb-constraint-definition — translates hardware requirements into layer stackup, netclasses, impedance matching, and custom DRC rules before layout starts.

• Layout & routing: pcb-layout-routing-automation — floorplanning, component placement, high-speed differential pair routing, thermal via arrays, and keepout enforcement.

• Validation & sign-off: pcb-validation-dfm-signoff — automated DRC/ERC, SI/PI screening, fab house DFM/DFA compliance, and production release sign-off (Gerbers/BOM/CPL).

• Schematic & datasheet auditing: schematic-datasheet-analysis — electrical rule auditing, datasheet grounding, pinout validation, and power tree tracing.

These ship in the separate @hybridlabor-api/bdb-hardware-pcb power-up module (KiCad + OpenSCAD MCP servers), not in this base skills package — install it as an optional AOS module when hardware work comes up.

---

🛡️ Overlap: The Supreme Godmodes

When do you use a godmode-* skill versus a narrower skill?

• godmode-ui-ux, godmode-engineering, godmode-shipping, godmode-3d-creation, godmode-media-creation, godmode-eventtech are orchestrator/enforcer skills. They hold the "supreme rulebook" for a domain. Use them when starting a major feature or defining the overarching architecture of a task.
• Use narrower skills (e.g., shadcn, drizzle-orm-expert, threejs-skills) for the tactical implementation inside those domains.

---

🛑 Fallback Note

If none of the intents above match your task, look for the closest general library skill in the catalog.

Remember: This file helps discover skills, but CLAUDE.md is the canonical source of truth for the primary Skill Routing by Domain table and the exact rules for Agent Counts (subagents vs teams vs solo). Do not rely on ask-tim to dictate pipeline definitions—always defer to CLAUDE.md.