agentleFS
cenconq25cenconq25/claude-code-app-studio/.claude/skills/start/SKILL.md

start

Skill40 starsChanged 7 months agoMITView on GitHub

First-time onboarding for the mobile app studio template. Asks where the user is starting from, classifies them, and routes them to the right next skill. Use this on the very first session, or whenever the project feels unmoored.

130 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use setup-engine, team-audio and 3 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • CoralGameSkill

    setup-engine

    Configure the project's game engine and version. Pins the engine in CLAUDE.md, detects knowledge gaps, and populates engine reference docs via WebSearch when the version is beyond the LLM's training data.

    Also in 13 projects that use this

  • CoralGameSkill

    team-audio

    Orchestrate audio team: audio-director + sound-designer + technical-artist + gameplay-programmer for full audio pipeline from direction to implementation.

    Also in 13 projects that use this

  • +3 more from CoralGame

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. 0 are common, 6 are distinctive, and 10 are very specific to this project.

  • 0

    Common

    Most of the 117 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 10

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not run when the project is mid-sprint and the user has a clear active task — /help or /sprint-status are better.

    line 25
  2. Distinctive:

    Do this without narrating.

    line 44
  3. Very specific:

    Framework configured?: Read .claude/docs/technical-preferences.md.

    line 47
  4. Very specific:

    Use AskUserQuestion once. Do not preload assumptions.

    line 60
  5. Distinctive:

    Wait for their pick. Do not proceed otherwise.

    line 69
  6. Distinctive:

    Ask them to share the theme in a sentence or two (free text).

    line 101
  7. Very specific:

    Ask them to give the one-liner (free text — open response, not AskUserQuestion).

    line 108
  8. Very specific:

    Check if production/review-mode.txt already exists.

    line 138
  9. Very specific:

    If it exists: Read and report — "Review mode is set to [value]." Move to Phase 4.

    line 140
  10. Very specific:

    If it does not exist: Use AskUserQuestion:

    line 142
  11. Very specific:

    Write the choice to production/review-mode.txt immediately:

    line 149
  12. Very specific:

    Create the production/ directory first if it is missing.

    line 154
  13. Very specific:

    Use AskUserQuestion to confirm the recommended first step.

    line 160
  14. Distinctive:

    If the user picks the alternative, accept their input and adjust.

    line 167
  15. Distinctive:

    Ask first; never assume the user's state.

    line 208
  16. Distinctive:

    If the user's situation does not fit a tab, listen and adapt.

    line 211

“Common” is measured: kept by at least 6 in 10 of the 117 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Guided Onboarding (Mobile App Studio)

Single file written by this skill: production/review-mode.txt (only when the user picks a review mode in Phase 3b).

This is the entry point. It assumes nothing — not the framework, not the idea, not the user's experience level. It looks at what is on disk, asks the user where they are, then hands off.


Purpose / When to Run

Run when:

  • The repo has just been cloned and nothing has been authored yet
  • The user types /start to re-orient
  • A returning collaborator wants a clean route into the right next step

Do not run when the project is mid-sprint and the user has a clear active task — /help or /sprint-status are better.

Inputs

  • .claude/docs/technical-preferences.md — to detect whether a framework has been pinned
  • design/concept.md — concept doc presence
  • design/prd/ — PRD count
  • production/review-mode.txt — existing review-mode setting
  • production/sprints/ and production/epics/ — production-stage signal
  • src/, ios/, android/, lib/ — code presence

Outputs

  • production/review-mode.txt — only written if not present and the user makes a choice in Phase 3b

Phase 1: Silent State Probe

Do this without narrating. The findings only inform recommendations; they are not the conversation opener.

Check:

  • Framework configured? Read .claude/docs/technical-preferences.md. If the Framework field still contains [TO BE CONFIGURED], the framework is not pinned.
  • Concept exists? Glob for design/concept.md.
  • PRDs exist? Count files matching design/prd/*.md.
  • Source code present? Glob for any of: src/**/*.ts, src/**/*.tsx, lib/**/*.dart, ios/**/*.swift, app/src/main/**/*.kt, **/*.swift excluding Pods.
  • Production work in flight? Count files in production/sprints/ and production/epics/.
  • Review mode set? Glob production/review-mode.txt.

Hold these as facts to validate the user's self-assessment in Phase 4.


Phase 2: Ask Where the User Is

Use AskUserQuestion once. Do not preload assumptions.

  • Prompt: "Welcome to the mobile-app studio template. Before recommending anything, where are you with the app right now?"
  • Options:
    • A) No idea yet — I want to figure out what to build.
    • B) Loose theme — I have a fuzzy sense (e.g., "a tracker for runners", "a calmer Instagram") but no spec.
    • C) Clear pitch — I can give you a one-liner: target user, core job, rough mechanic.
    • D) Existing app — There is already code, screens, or planning. I want to organize it.

Wait for their pick. Do not proceed otherwise.


Phase 3: Route Based on Answer

If A: No idea yet

  1. Acknowledge that starting cold is normal.
  2. Recommend /brainstorm open to develop a concept from zero. Mention /brainstorm <hint> if even a single seed word comes to mind.
  3. Show the path: Concept stage
    • /brainstorm open — produce design/concept.md
    • /setup-framework — pin RN / Flutter / iOS / Android
    • /design-bible — visual identity tokens, app icon, splash
    • /map-systems — break the concept into systems
    • /design-system <system> — one PRD per system
    • /consistency-check and /review-all-prds Architecture stage
    • /create-architecture — master architecture
    • /architecture-decision (×N) — ADRs from the required list
    • /create-control-manifest
    • /architecture-review Pre-production
    • /ux-design — per-screen UX specs
    • /prototype — validate the riskiest mechanic
    • /create-epics then /create-stories
    • /sprint-plan — first sprint Production — pick stories with implementation skills (in the second skill set).

If B: Loose theme

  1. Ask them to share the theme in a sentence or two (free text).
  2. Validate it as a starting point — never redirect.
  3. Recommend /brainstorm <hint>.
  4. Show the same path as A, with the brainstorm step substituted.

If C: Clear pitch

  1. Ask them to give the one-liner (free text — open response, not AskUserQuestion).
  2. Use AskUserQuestion:
    • Prompt: "How do you want to proceed?"
    • Options:
      • Formalize first — Run /brainstorm <pitch> to structure it into design/concept.md.
      • Skip ahead — Go straight to /setup-framework, then author the concept doc by hand.
  3. Show the path:
    • /brainstorm or /setup-framework (their choice)
    • /design-bible after the concept doc exists
    • /design-review design/concept.md
    • /map-systems
    • then PRDs → architecture → epics as in A.

If D: Existing app

  1. Reflect what you found in Phase 1, plainly: "I see [X PRDs / Y source files / framework set to Z / no concept doc]."
  2. Sub-case D1 (early — only a concept or only a framework pin): recommend /setup-framework if missing, then /project-stage-detect.
  3. Sub-case D2 (significant artifacts — PRDs, ADRs, stories already present):
    • Explain: "Files existing is not the same as the template's skills being able to read them. PRDs may be missing required sections; ADRs may lack Status. /adopt checks compliance specifically."
    • Recommend in order:
      1. /project-stage-detect — what stage and what is missing entirely
      2. /adopt — format compliance and migration plan
      3. /setup-framework if not pinned
      4. /reverse-document for any code with no design doc behind it
      5. /architecture-review to bootstrap the requirements registry

Phase 3b: Set Review Mode

Check if production/review-mode.txt already exists.

If it exists: Read and report — "Review mode is set to [value]." Move to Phase 4. Do not re-ask.

If it does not exist: Use AskUserQuestion:

  • Prompt: "How much review do you want as you move through the workflow?"
  • Options:
    • Full — Director-tier reviews at every gate. Best for teams or anyone learning the template.
    • Lean (recommended) — Reviews only at phase boundaries. Balanced for solo and small teams.
    • Solo — No director reviews. Maximum speed. For prototypes, hackathons, or experienced solo devs.

Write the choice to production/review-mode.txt immediately:

  • Full → full
  • Lean (recommended) → lean
  • Solo → solo

Create the production/ directory first if it is missing.


Phase 4: Confirm Before Handing Off

Use AskUserQuestion to confirm the recommended first step. Do not run another skill yourself.

  • Prompt: "Want to start with [recommended skill]?"
  • Options:
    • Yes — start with [skill]
    • Something else first

If the user picks the alternative, accept their input and adjust.


Phase 5: Hand Off

Respond with one short line: "Type [skill command] to begin." Do not re-explain or pad.

Verdict: COMPLETE — user oriented and routed.


Edge Cases

  • User picks D but disk is empty: "The template looks unused — A or B may fit better. Which is closer?"
  • User picks A but disk has working code: "I noticed code in [path]. Did you mean D?"
  • Returning user, fully set up: Skip the long flow — "Framework: [X]. Concept: present. Review mode: [value or 'lean (default)']. Want to continue with /sprint-plan, /sprint-status, or describe what you want?"
  • None of the options fit: Let the user describe their state in free text and adapt.

Examples

User picks B) Loose theme, types "an app that helps bouldering gym climbers track their projects":

  • Skill recommends /brainstorm bouldering project tracker.
  • Shows the full concept → architecture → production path.
  • Asks for review mode (user picks Lean).
  • Writes production/review-mode.txt with lean.
  • Confirms first step, hands off.

User picks D) Existing app, project has 3 PRDs and an iOS Xcode project:

  • Reports findings.
  • Recommends /project-stage-detect then /adopt.
  • Notes that /setup-framework may not be needed if Xcode metadata already implies iOS native.
  • Asks for review mode if unset.
  • Hands off to /project-stage-detect.

Collaborative Protocol

  1. Ask first; never assume the user's state.
  2. Present options as choices, not orders.
  3. Recommend, do not auto-execute.
  4. If the user's situation does not fit a tab, listen and adapt.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.