Council Review

Canonical source: docs/council/COUNCIL-REVIEW.md · Snapshot commit: dbd497b496c0

Life in Days — Product Council planning-baseline review

Review date: 2026-08-13 Decision: G0 planning baseline approved Next gate: G1 shared-understanding confirmation by Arun Evidence boundary: documentation and planning review only; no implementation, model evaluation, credential collection, provider configuration, infrastructure mutation, deployment, recovery exercise, or production verification is claimed

Reviewed artifacts

• Product Council charter and four role charters
• Product Requirements Document
• UX Specification
• Project Tracker
• Requirements Traceability
• Implementation Plan
• The discovery requirements, shared-understanding proposal, domain language, research, AI evaluation protocols, and storage evaluation referenced by those artifacts

Council participants and final decisions

Council seat	Review focus	Final decision
Senior Product Manager	Scope, all 78 requirement definitions, acceptance behavior, deferrals, and requirement-level traceability	Approved — no remaining P0/P1 product blocker
UI/UX Design Lead	Complete user-visible behavior, accessibility, privacy cues, state recovery, and canonical gate flow	Approved — no remaining P0/P1 UX blocker
Technical Architect	Feasibility, data durability, external side effects, privacy enforcement, export, backup, migration, and recovery contracts	Approved — all nine initial architecture findings resolved; no new P0/P1 blocker
Project Manager	Milestones, epics, task dependencies, evidence, status accuracy, and G0–G9 sequencing	Approved — no remaining P0/P1 planning blocker; dependency-cycle audit passed
Product chair	Cross-artifact reconciliation, mechanical validation, evidence boundary, and gate decision	Approved — G0 closed; G1 remains owner-controlled and blocked

Material findings resolved during review

1. Added a durable Needs Date Review holding model with nullable Journal Day only in that state and acknowledgement only after encrypted durable preservation.
2. Standardized every governing artifact on one G0–G9 gate sequence and corrected the critical-path diagram so Recovery Ceremony G8 precedes private-launch acceptance G9.
3. Distinguished same-day artwork staleness from redating/source removal: same-day changes retain visibly stale art; art leaves active presentation only when a bound source no longer belongs to that day.
4. Preserved the approved Artwork Suppression rule: it blocks only automatic sweep recreation; a manual request remains available under normal gates without silently clearing suppression.
5. Added an export ADR contract for ephemeral passphrase handoff, restart failure, partial cleanup, single-use download leasing, and server-observable successful-stream completion.
6. Removed browser credential entry; the browser exposes only configuration/health state while secrets remain runtime/server-side.
7. Added optional owner-authored private image descriptions for accessible real-photo text alternatives without sending photo data to AI.
8. Preserved one immutable VoiceNotes Integration Activation instant; no re-enable epoch can create an automatic historical-import path.
9. Added external-provider prepared/sending/unknown_outcome/confirmed states so ambiguous AI calls are not blindly replayed and conservative spend remains reserved.
10. Added a durable capture-intent/reconciliation design across database and filesystem writes, plus a bounded durable Telegram album-settling algorithm.
11. Added an executable, fail-closed R2-to-Restic source design using an application-consistent manifest and read-only mount or streaming filesystem.
12. Replaced invalid family-level traceability with a 78-row matrix using only real PRD, UX, tracker, and architecture identifiers.
13. Removed five tracker dependency cycles by separating architecture/host/privacy/domain contracts from their downstream implementations and verification tasks.

Validation evidence

• Canonical PRD definitions: 78.
• Requirements traceability rows: 78, with zero missing, extra, or duplicate requirement IDs.
• Matrix references validated: all referenced UX and tracker IDs resolve.
• Tracker dependency audit: zero dependency cycles and zero unknown task references among executable/deferred tracker IDs.
• All local Markdown link targets resolve after the final index update.
• git diff --check passes.
• Council approvals above are planning judgments, not usability-test, security-test, implementation, restore, deployment, or production evidence.

Gate disposition

G0 is complete because the requested council artifacts exist, their material contradictions are closed, and all four specialist seats approved the integrated baseline. G1 is not complete: Arun must explicitly confirm the proposed shared understanding. That confirmation authorizes only the next separately gated work. It does not authorize credentials, paid actions, provider/account changes, application implementation, DNS/server changes, deployment, or launch.