idea-to-appstore-autopilot

name	idea-to-appstore-autopilot
description	Use when the user wants autonomous end-to-end mobile app delivery from idea through implementation, verification, TestFlight upload, and App Store review submission with minimal human intervention.

Idea To App Store Autopilot

Overview

Run OMO as an autonomous delivery system for Expo apps. Use Sisyphus plus /ulw-loop for long-run persistence and loop continuation. Use Hephaestus as a deep worker for difficult implementation tasks inside the loop.

Hephaestus alone is not the loop controller. For multi-hour autonomy, keep /ulw-loop enabled.

Operating Mode

1) Controller Choice

• Default controller: Sisyphus
• Deep execution bursts: delegate to Hephaestus
• Planning and hard tradeoffs: delegate to oracle

2) Submission Policy

• Always use API/CLI first for store operations.
• Use browser automation only as fallback for steps not reliably exposed by API/CLI.
• Never use browser-first for everything.

3) Context Rot Prevention

• Run in cycles of 45-90 minutes.
• At cycle end, persist state artifacts and start next cycle from artifacts, not transcript memory.
• Keep current cycle scope small: one milestone at a time.

Required Artifacts

Maintain these files during execution:

• .autonomy/MISSION.md
• .autonomy/STATE.json
• .autonomy/BACKLOG.md
• .autonomy/DECISIONS.md
• .autonomy/EVIDENCE/<cycle-id>.md
• .autonomy/RELEASE.md

End-To-End Workflow

Phase 0: Preflight

1. Verify secrets and credentials.
2. Verify build and submission toolchain.
3. Verify app identifiers and signing setup.
4. Refuse to proceed only on true hard blockers (missing credentials, missing app identifiers, revoked access).

Use references/checklists.md preflight list.

Phase 1: Market Intelligence (Autonomous)

1. Research competitors, reviews, Reddit/X/web complaints, and feature gaps.
2. Build a ranked opportunity backlog:
   • complaint frequency
   • competitor weakness severity
   • implementation effort
   • expected user impact
3. Pick top differentiators to implement first.

Phase 2: Product + Technical Plan

1. Convert idea + research into PRD and milestone plan.
2. Define acceptance criteria for each milestone.
3. Define verification evidence required per milestone.

Phase 3: Build Loop (Autonomous)

1. Execute milestone tasks with parallel subagents for independent lanes.
2. Merge and run integration pass.
3. Persist artifacts.
4. Continue next milestone automatically.

Phase 4: Verification Gates

Run all gates before release steps:

1. Lint/typecheck/static checks.
2. Unit and integration tests.
3. UI/navigation checks:
   • Playwright for web/admin/user web surfaces.
   • Native smoke tests for core mobile flows.
4. Regression checks on critical flows (auth, onboarding, paywall, purchase, profile/settings).

Phase 5: Distribution and Review Submission

1. Build iOS release with EAS.
2. Submit to TestFlight with EAS Submit.
3. Complete App Store metadata/review fields through API/CLI.
4. If any required field is blocked by API limits, use Playwright browser fallback.
5. Submit app for App Review.

Use references/runbook-commands.md for command templates.

Completion Criteria

Mission is complete only when all are true:

1. Requested feature/app scope is implemented.
2. Verification gates pass.
3. Build is uploaded and visible in TestFlight processing or ready-to-test.
4. App Store review submission is sent.
5. Evidence is logged in .autonomy/EVIDENCE and .autonomy/RELEASE.md.

Safety Rules

• Do not delete production data.
• Do not release to public production automatically unless user asked.
• Do not bypass verification gates.
• Do not claim completion without evidence artifacts.

Mission Prompt

Read and use references/mission-template.md as the default /ulw-loop mission prompt.