WorldChest — evidence programme (graduated from incubator #5)

[don-petry]

NOT AN OFFICIAL MINECRAFT PRODUCT. NOT APPROVED BY OR ASSOCIATED WITH MOJANG OR MICROSOFT.

This Discussion carries WorldChest's approved incubator idea into its product repo so the org planning pipeline can plan from it. It condenses the PRD; the PRD is the source of truth.

• Source idea: combsmith/incubator Discussion #5 (working name BlockVault; brand decision 2026-10-19). Approved there with idea:approved on 2026-10-21.
• Planning artifacts: _bmad-output/planning-artifacts/: prd.md (John, headless bmad-prd), brief.md, market-research.md, brainstorm.md, ost.md.
• Incubator record: combsmith/incubator ideas/worldchest/ (PR #184); seeded here in PR docs: seed planning artifacts from the incubator package (graduation) #5.

Scope (PRD scope note)

The brief recommends park: buy evidence first. The deliverable is therefore a one-week evidence programme, not the product: brief §6 tests 1–4 plus a time-boxed incumbent watch (test 5). Nothing below authorises product code, an Android or iOS app, a backend, a store listing or a paid plan. The backup product appears only as the hypothesis under test. A product PRD is a later update of prd.md, written after the results exist.

Hypothesis under test (not authorised to build): tiered access sold to parents as a Family plan (about $5/mo, an assumption): a zero-setup .mcworld share-sheet baseline, a Shizuku "power mode" for automatic versioned whole-library backup and restore using the POC's include/skip file set, a desktop companion for big migrations, and the official Mojang API as the long-term goal.

Functional requirements (all evidence-gathering)

ID	Requirement	Brief §6 row
FR-1	Fake door on worldchest.app: a Family-plan page at the tested price, with the Naming rule and Disclaimer, an adult parent/guardian confirmation before the email field, no payment, storing only an email and timestamp per sign-up plus minimal anonymous session analytics. Conversion threshold, visitor floor, test window, analytics provider, retention and privacy wording are dated before the first visitor. Visitors, sign-ups and rate recorded.	1 Demand
FR-2	Five parent interviews, recruited regardless of incident history; each dated record notes whether a lost-world or device-change incident happened in the last 12 months, what they did, and what backup-like services they pay for. No children interviewed.	1 Demand
FR-3	Shizuku hallway test: one onboarding script, three non-technical parents on their own Android 11+ phones, unaided. End state is Shizuku reporting its service running. Record completion, computer-free vs computer-assisted, elapsed time, the step where they stopped, and device model, OEM, Android and Shizuku versions. Phone restored to its original settings afterwards.	2 Onboarding
FR-4	Share-sheet round trip: export one world as .mcworld and re-import it on one Android 15 device and one iOS device; record whether export exists, where the file lands, whether re-import loads, plus game and OS versions. One-day time-box; unfinished = "not shown".	3 Platform
FR-5	EULA and Usage Guidelines read: a dated one-page note quoting the clauses relied on, with a verdict of permitted / prohibited / unclear and what would change it; a written question sent to Mojang brand or legal, with send date and any reply.	4 Legal
FR-6	Brand compliance carried forward: everything the programme publishes follows the Naming rule ("Minecraft" only as a secondary title) and carries the Disclaimer.	4 Legal
FR-7	Mojang native-backup watch: dated checks of the feedback site and release notes, at least once during the programme and once before the decision. A native feature is a "no" only if it closes the core gap (automatic whole-library backup with versioned restore).	5 Incumbent
FR-8	Results into the brief and this record: each of brief §6 rows 1–5 gets a dated result ("not shown" or "not run" with a reason allowed); a summary comment lists results against graduation criteria (a) and (b) and the "say no" conditions. Criterion (c), an end-to-end restore by a non-technical user, needs product code and is reported "not testable in this programme".	Decision

Success metrics

Primary

• SM-1 Incident rate in interviews: at least 3 of 5 parents describe a lost-world or device-change incident (FR-2).
• SM-2 Fake-door conversion: at or above the pre-set threshold, and only if visitors reach the pre-set floor; below the floor it is inconclusive (FR-1).
• SM-3 Shizuku setup feasibility: at least 2 of 3 parents reach the end state computer-free and unaided (FR-3).
• SM-4 Legal verdict recorded and "permitted". "Prohibited" is a no; "unclear" blocks approval (FR-5).

Secondary

• SM-5 Share-sheet round trip succeeds on both devices (FR-4); the brief sets no target.
• SM-6 About five working days from first test to decision comment (FR-8).

Counter-metrics (do not optimise): waitlist size bought with urgency copy; any line of product code written; help given during a hallway test.

Guardrails and non-goals

No child is a participant and no child data is collected. No world files are uploaded. About one week of the owner's time, no paid traffic, no counsel. Explicitly out: product code, apps, backend, store listing, Google Play policy testing, payments, versioning/rewind spike, end-to-end restore test, desktop companion, creators/streamers/classrooms, Java Edition.

Planning note

Stories planned from this Discussion should be the eight FRs above as evidence tasks with their testable consequences from prd.md §4, ordered so that FR-1's thresholds are dated before launch and FR-8 comes last. They are not build tasks.

[don-petry]

📋 Initiative planned by the BMAD Scrum Master (Bob).

Epic #7 — WorldChest — one-week evidence programme (brief §6 tests 1–5, decision return)

8 stories created (inert — labelled initiative, NOT initiative:auto):

• [Phase 1] FR-6: Brand-compliance checklist for everything the programme publishes #8 (S) — [Phase 1] FR-6: Brand-compliance checklist for everything the programme publishes
• [Phase 2] FR-5: EULA and Usage Guidelines note + written question to Mojang #9 (S) — [Phase 2] FR-5: EULA and Usage Guidelines note + written question to Mojang
• [Phase 2] FR-1: Fake door on worldchest.app — pre-register thresholds, launch, record conversion #10 (M) — [Phase 2] FR-1: Fake door on worldchest.app — pre-register thresholds, launch, record conversion
• [Phase 2] FR-2: Five parent interviews, recruited regardless of incident history #11 (M) — [Phase 2] FR-2: Five parent interviews, recruited regardless of incident history
• [Phase 2] FR-3: Shizuku hallway test — three non-technical parents, unaided #12 (M) — [Phase 2] FR-3: Shizuku hallway test — three non-technical parents, unaided
• [Phase 2] FR-4: Share-sheet .mcworld round trip on Android 15 and iOS (one-day time-box) #13 (S) — [Phase 2] FR-4: Share-sheet .mcworld round trip on Android 15 and iOS (one-day time-box)
• [Phase 2] FR-7: Mojang native-backup watch (time-boxed, dated checks) #14 (S) — [Phase 2] FR-7: Mojang native-backup watch (time-boxed, dated checks)
• [Phase 3] FR-8: Results into brief §6 and the decision comment #15 (S) — [Phase 3] FR-8: Results into brief §6 and the decision comment

Open questions for review:

• Which Discussion carries the FR-8 summary and decision: incubator Discussion docs: seed planning artifacts from the incubator package (graduation) #5 (as the PRD says) or worldchest Discussion WorldChest — evidence programme (graduated from incubator #5) #6? Incubator docs: seed planning artifacts from the incubator package (graduation) #5 already carries idea:approved (2026-10-21), which the PRD reserves for the post-results decision — how should that label be read?
• Where in combsmith/worldchest (public) should programme working docs live — brand checklist, EULA note, Onboarding script, watch note? No docs/ tree exists; only _bmad-output/planning-artifacts/. Raw participant data must stay out of the repo regardless.
• PRD §9 Q2/Q3: recruitment channel and interviewer for FR-2/FR-3, and which iOS device/version is used for FR-4 — to be fixed by the Decider before those tests start.
• [cost_cap] Interview compensation is left to the Decider with no upper bound (PRD FR-2). The Decider should set a ceiling, or confirm $0, before recruiting.

---

Review the epic and its sub-issue DAG, adjust as needed, then add initiative:auto to epic #7 to hand it to initiative-driver for auto-implementation.

[don-petry]

Owner decisions on the planner's open questions (2026-10-22)

• Decision venue: results and the build / close / re-park decision are posted on worldchest Discussion WorldChest — evidence programme (graduated from incubator #5) #6. idea:approved on incubator Discussion docs: seed planning artifacts from the incubator package (graduation) #5 means "graduate and run the evidence programme". Applies to [Phase 3] FR-8: Results into brief §6 and the decision comment #15.
• Working docs: _bmad-output/evidence/, one file per story; no raw participant data in this public repo. Applies to [Phase 1] FR-6: Brand-compliance checklist for everything the programme publishes #8, [Phase 2] FR-5: EULA and Usage Guidelines note + written question to Mojang #9, [Phase 2] FR-3: Shizuku hallway test — three non-technical parents, unaided #12, [Phase 2] FR-7: Mojang native-backup watch (time-boxed, dated checks) #14.
• Interviews ([Phase 2] FR-2: Five parent interviews, recruited regardless of incident history #11): owner's own network, owner interviews, $0 compensation.
• Devices and roles ([Phase 2] FR-3: Shizuku hallway test — three non-technical parents, unaided #12, [Phase 2] FR-4: Share-sheet .mcworld round trip on Android 15 and iOS (one-day time-box) #13): household iPad or iPhone for FR-4; the owner runs every test.

PRD updated in #16.