Known Issues / Open Items
Tracked here so nothing has to be rediscovered by stumbling into it.
Active
(nothing currently blocking)
Known Limitations (by design, not bugs: noting so they’re not mistaken for gaps)
- The prototype is mostly a scripted walkthrough. Three things are genuinely interactive: the breathing exercise (Flight Attendant and Pilot, panel 4) runs a real timed session you can pause and resume; the “Breathe & Box” button on every persona opens a working Box Breathing tool (4-4-4-4) with a short definition; and the Wellness Toolkit cards respond to taps with a “launches in Phase 2” note. The highlighted option on each panel 3 advances the story like the arrow does. The other option buttons are not wired. Everything else is narrative, and nothing is wired to real audio/video content, a backend, or real data. Expected for a portfolio concept piece, not a shipped product.
- Analytics on the demo page (Microsoft Clarity) will only reflect a handful of visitors, so any observation drawn from it is anecdotal, not statistically meaningful. The app concept’s zero-data promise also means true DAU/retention tracking is deliberately not possible; usage would be measured as aggregate event counts only.
- Wireframes (
wireframes/index.html) currently cover only the Flight Attendant persona’s core flow. The Reserve Crew and Pilot storyboards exist in the high-fidelity prototype but don’t have a corresponding low-fi wireframe pass.
- Tested in Chromium only. Not verified on Safari/Firefox or real mobile devices, likely fine given it’s plain CSS/JS with no exotic APIs, but not confirmed.
Candidate Improvements (open decisions, not bugs)
- AI/automation framing: upgrade the bulletin board’s “automated keyword/toxicity filter” (currently US-303 in the backlog) from a plain keyword list to an on-device ML toxicity classifier, framed explicitly as a deliberate PO scoping decision: real AI/automation substance, zero conflict with the Zero-Data Privacy Vault since it classifies message content, not user behavior. Agreed in discussion to be the stronger AI feature over a personalization/”pin your most-used button” idea, which was considered and rejected as thin/checkbox-AI. Not yet written into the docs. (The roadmap and backlog now agree that US-303 is a keyword/toxicity filter, so this is purely an upgrade decision, not a correction.)
- Wireframes: persona coverage gap. Only the Flight Attendant flow has a wireframe pass; Reserve Crew and Pilot personas exist in the high-fidelity prototype but not in wireframe form.
Active: Found in 2026-10-13 Full-Project Scrutiny Pass
Reviewed as a skeptical PO would, looking for exactly what a sharp interviewer might catch. Ranked by severity. Four of the five were fixed on 2026-10-14 (see Resolved); only the push-notification question below remains open.
Open: Push notifications vs. the privacy promise, unaddressed. OKRs target “<10 min average time-to-first-response” and “30%+ weekly engagement” on the bulletin board, but nothing in the backlog explains how a user learns a new post/reply exists without a push notification, and push requires a device token registered with Apple/Google, in tension with US-201’s “no device identifier is ever transmitted” promise. No story or doc currently resolves this.
Judgment calls (not errors, but worth being ready to defend): the 30%-engagement-in-first-week OKR target may be aggressive for a product whose core use case is rare acute-panic events, not daily engagement; and publicly shipping KNOWN_ISSUES.md itself is a deliberate transparency choice that could read as unfinished to a non-technical reviewer, even though it reads as mature practice to a technical one.
Parked Ideas (logged 2026-10-14, not scheduled)
- Data-model choice. The concept currently promises zero data (option A). Alternatives: B, an anonymous install ID with measured aggregate usage; C, identified accounts. B is the likely middle path if measurement becomes a priority. Decision deferred.
- Substance-use support (“Clear Skies”-style resources). Many airlines run peer programs for this, so a resources panel may belong in the app. It conflicts with the roadmap’s “no company-integrated EAP referral” rule, and the details of any specific airline program are unverified. Needs a deliberate scope decision.
- Sample workouts and functional option buttons. Most option buttons are not wired and there is no workout content yet. Adding short sample sessions and making the buttons navigate would deepen the prototype.
- One-page Measurement Plan. Event taxonomy and anonymous counts for engagement metrics (sections opened, exercises started and completed).
- Early validation. Talk to 2 or 3 former crewmates and add an “Early validation” section to the README.
Resolved
Dead cross-reference in roadmap.md (2026-10-14): now points to the Phase 2 table in mvp-scope.md, where the reasoning actually lives.roadmap.md overclaimed an on-device moderation classifierUS-102 cold-start acceptance criterion (500ms) was unrealistic (2026-10-14): split into a 500ms tap-to-start target once the app is open and a under-3-second cold-launch target on a mid-range device.US-301 and US-303 described moderation separately (2026-10-14): now one two-stage pipeline (on-device keyword check, then toxicity filter), cross-referenced in both stories.Prototype had no real interaction to measure (2026-10-14): breathing exercise is now a working timed session (Flight Attendant and Pilot), toolkit cards are tappable, and the demo page is instrumented with Microsoft Clarity.Stale “Product Owner, Reservations” references (2026-10-14): footer, README, okrs.md and product-vision.md reworded so nothing is tied to one specific posting. (An earlier pass missed the docs/ folder; caught in a later review.)“Nothing recorded” wording could read as contradicting the demo page’s analytics (2026-10-14): scoped the claim to the app, and added a disclosure to the footer and README.GitHub Pages not enabled, live prototype link 404’d: enabled 2026-10-12, confirmed live at serrayon.github.io/delta-wellness-app/prototype/Wireframes: style/depth decision (decided 2026-10-12): keep the plain low-fidelity box style as-is, no restyle or annotations needed.Definition of Done doc: written, see docs/definition-of-done.mdStakeholder map / roadmap sequencing doc: written as two separate docs: docs/stakeholder-map.md and docs/roadmap.md