TaleWater — Build Dashboard

Generated 2026-05-29 10:33 PT · branch main · commit f460f5b
24slices closed
3open gaps
7open questions
6open audit findings

Active

Track A Proposed by Claude 2026-05-29
Track B   No active slice on Track B.

Up next — Nate’s priority ranking

1. Dead code & asset sweep
`src/app/explore.tsx`, web variants, retired splash HTML, `assets/expo.icon/`. Smallest entry on this list; clears tech debt before it ossifies.
2. DAL + migration tests against real SQLite
foundational. Every future `src/db/` slice gets cheaper after this lands.
3. Drag-to-place on the S-15 placement modal
UX gap surfaced *during* S-15 acceptance. Same DAL call; scoped MapLibre swap.
4. Manual "retry weather" affordance
small UX gap left by S-16; covers outage recovery and pin-move re-fetch.
5. Session-edit surface
§4.3 names title / water-temp / clarity / flow as editable but no screen sites the affordance. Candidates: screen 11 (session picker) or screen 07 (observation detail's session-conditions block).

V1 — track roadmap (from SEQUENCE.md)

Track A   native app
1. Native shell (scaffold + launch experience — `SPEC.md` §7.0 — complete as of S-02) + SQLite + the migration runner. The SQLite layer is the next sub-step.
2. Transcript upload → extraction → provisional records → session-scoped log, on a **mock extraction response**. Q-SE1 resolves within this step. Screens 02, 04, 05, 06, 07, 11.
3. Map + pins. Screens 01, 03.
4. GPS session + reconciliation. Depends on Q-G1. Screens 08, 09.
5. Remaining polish
weather auto-fill, detail-view polish, filters, screen 10. Cuttable before slipping the core loop.
Track B   extraction
1. The extraction prompt, tuned against the real sample transcript in a throwaway harness. Strategy captured in `EXTRACTION.md` — categories, principles, field shapes, and pipeline architecture. The high...
est novel *product* risk — distinct from the highest *execution* risk, which is shipping itself (`SPEC.md` §11 R0, `CLAUDE.md` goal section).
2. The extraction proxy. Depends on Q-A1.

V2 — Intended-section product order

1. §14 Email ingestion
removes the manual file-pick step; also lays the backend groundwork §18 auth extends.
2. §15 Offline map tiles
North American backcountry often has no signal.
3. §16 Insights & spatial heatmap
needs accumulated observations to be meaningful. May absorb the V2 computed-pattern Diary entries from the 2026-05-26 BACKLOG entry.
4. §13 Photo intelligence
photo library scan, Claude Vision, compound extraction merging photo + transcript.
5. §17 Export
CSV / shareable summary.
6. §18 Multi-user & sharing
requires location encryption.
7. §19 Manual entry
the un-recorded-session input path. Low priority by product call; sequenced last.

Punted — open gaps (V1 work not yet delivered)

G-02   Scaffold coordinates on observations (no real §5.2 / §6.6 pipeline)   Partially closed

Status
Partially closed by S-14 on 2026-05-25 (reconciliation half); trip-pin half open.

Expected next slice
§6.6 (trip-pin fallback) closes the remaining half.

G-05   Delete and pin-drag affordances missing on the detail screen   Partially closed

Status
Partially closed by S-15 on 2026-05-25 (pin-adjust half); delete half open.

Expected next slice
TBD — pin-drag pairs with the embedded-map

G-07   Diary Fisheries axis still reads `diary_entries.fisheries` strings, not `fisheries.id` FKs   Open

Status
Open

Expected next slice
Diary edit-affordances slice (post-S-21).

Open questions

Technical Q-A1 Extraction proxy provider + abuse control
Cloudflare Workers vs. Vercel for the stateless extraction proxy,
Technical Q-S1 Sentiment inference reliability
Can sentiment be reliably inferred from oblique transcript lines
Technical Q-G1 Background GPS lifecycle (narrowed by S-21)
Behavior on **mid-session permission downgrade** — a user toggles
Technical Q-M1 Offline tile strategy
How offline map tiles are stored for North American backcountry use.
Product Q1 Ground-truth labeling feasibility
Is it feasible for Nate to manually label ~10 real transcripts
Product Q3 iOS background-location battery drain
The real-world magnitude of background-location battery drain.
Product Q4 Offline tile strategy (North American backcountry signal)
How the app behaves with no cell signal in North American backcountry —

Open audit findings

A-04   M1 (extraction capture rate) is unmeasurable as designed   Open

The hole
`SPEC.md` §6.2 says raw transcripts are discarded after

Verdict
Open — kept Open. The §9 success metrics are acknowledged as

Re-raise condition
N/A — item is Open; reviewed weekly.

A-05   Proxy "daily ceiling" is not a real spend control   Deferred

The hole
§3.3 names the per-day request ceiling on the proxy as the

Verdict
Deferred. Hold until the proxy build slice.

Re-raise condition
Proxy build slice is proposed or starts.

A-08   Build-loop reconcile tax is unproven   Open

The hole
Every slice must touch up to four artifacts (`Completed_Slices`,

Verdict
Open — kept Open. The value of this kind of process investment

Re-raise condition
N/A — item is Open; reviewed weekly.

A-14   Zero retained tests after six slices   Open

The hole
Acceptance tests have been verified with `__DEV__`

Verdict
Open — regression tests are required. Two-part plan: (1) the

Re-raise condition
N/A — item is Open; reviewed weekly.

A-17   Hand-rolled migration runner is a footgun   Open

The hole
First-time builder, custom migration code, no tests,

Verdict
Open — keep the hand-rolled runner, add unit tests for it as

Re-raise condition
A migration step fails in production (user data

A-22   Proxy threat model is absent   Deferred

The hole
No auth between app and proxy, no per-IP rate limit, no

Verdict
Deferred. Same home as A-05 — address auth,

Re-raise condition
Proxy build slice is proposed or starts.

Recent closes

Track B S-24 extraction-prompt-v02Passed2026-05-29
Track A S-22 fishery-naming-and-waterway-autocompletePassed2026-05-29
Track A S-21 gps-background-safetyPassed2026-05-28
Track B S-20 extraction-harness-v01Passed2026-05-27
Track A S-19 fishery-entity-and-manual-assignmentPassed2026-05-28