Cosmo V2 / Information Architecture·M2 restructure
Cosmo V2 Information Architecture
Sunday 2026-05-10 review with Sergey and Lee. Operational restructure for M2: three top-level surfaces, one full-screen Settings, two shared Detail Panels. Observability deferred to M3 with 27 jobs defined per role.
2026-05-09 v0.1
M2 scopeM3 deferredDropped
01Hierarchy
Surface tree
Three slots in the main menu, plus Settings. Two shared Detail Panels mount as slide-overs from the main pages, with deep-link routes for power users. Role routing is provisional silo (validate with Elad before locking, AB-030).
Main menu / Workshop + Weekly are the two surfaces being designed
PG operational hub · designed now
Workshop
PG day-to-day work surface. Single unified playable-as-task list with a Pipeline Tracker top section, an Ideas side panel, a Feedback side panel, and a Pace observability strip. Detail Panel mounts from any row. Absorbs V1 Delivery, Ideation PG fragments, and Pipeline Tracker (V1 /tracking). Named "Workshop" 2026-05-09, PG-as-craftsperson framing.
New M2PG primaryTL cross-view/workshop
Top section · Pipeline Trackercross-system join · V1 /tracking absorbed
Monday
Auditor
Apollo
Build
Delivery
Main list · 5 named views
Needs actionAging watchSend to clientPer-game paceRevisions priority
Deep Ocean TreasuresBestPlay · Tatiana · revision · day 3
Disney Solitaire 0411SuperPlay · Lex · ready to send
Royal Match S07CrazyLabs · Daria · build failed
Wordscapes ArenaTripledot · Veronika · waiting on idea
Match-3 BanditCrazyLabs · Tatiana · TL approval
Pace observability strip
Royal Match42 / 60
Disney Solitaire70 / 60
Deep Ocean28 / 60
Detail Panel mounts from any row (slide-over)Open decisions: tracker default state · view shape · ideas placement · feedback affordance
Weekly
M1 launching
Mon-Fri kanban grid. Auditor rows, task cards, quota backlog right rail, manage-quotas modal. Absorbs Revisions backlog, per-auditor Monday view, build chips, carryover group. Partner surface to Workshop, both consume the same orchestration-layer state.
M2 add Carryover group in Quota Backlog (P70), unfinished items roll forward by default
M2 add Build chip on ply-cards, opens Build Detail drawer (status, retry, Apollo deep-link)
TL primaryPG read-mostly/weekly
Settings
5 sub-sections
Full-screen surface. Barely opened daily, can hold a lot. Admins write, Head-of-PG writes entities + allocations, TLs write own team allocations, PGs write owned games. Out of design focus for Sunday review.
Schedulesfolded into Capacity (vacation, sick, recurring availability are capacity overrides)
Importsa flow inside Settings:Games (CSV bulk import) or Settings:Admin, not its own sub-section
User prefslives in the profile menu (top-right avatar dropdown), not Settings nav
Monitoringsplit TBD per AB-081, decide post-Sunday
Detail panels / shared, two mount modes
Playable Detail Panel
slide-over · /task/:taskId
Mounts from Workshop (main list + Pipeline Tracker top section) and Weekly. Header + State history + Apollo/Build/Idea panels + Feedback thread + Revision status. Six IDs hidden by default in Technical details.
All rolesActive V2AB-083, AB-090, AB-078, AB-079
Game Details Panel
slide-over · /game/<id> · secondary
Mount via View Game from anywhere; standalone deep-link /game/<id>. Same content component, two modes. Routing decision post-Sunday (AB-097).
All rolesActive V2AB-095, AB-096, AB-097, AB-098
Deep-link routes / power users + admin
/task/:taskIdPer-task deep dive. Full orchestration + Apollo task + solution paths + state timeline. Reset / Cancel and Reuse / Archive / Bug Report. From Workshop tracker rows.V1 surface
/orchestration/:orchestrationIdSame component as /task. Resolved by which ID param is present. Used for support tickets, debugging, deep-link sharing. State timeline = reference implementation for Detail Panel history (AB-083).V1 surface
/game/<id>Full-page Game Details, same content as slide-over. Standalone, deep-linkable, shareable. For support, leadership, cross-team. Decide post-Sunday (AB-097).Proposed
12 V1 surfaces dropped or foldedexpand for table
V1 surface
Reason dropped
Disposition
Status
Observability only. Currently broken (P1, P34).
Two-bar pace pattern → PG Op observability strip. Full charts → M3.
Dashboard
Observability only. Near-empty body, no clear primary user.
Aging-table fragment → PG Op Aging view. Trend chart → M3.
Ideation
No coherent purpose in V2. Fragments redistribute. Send-to-Apollo automated.
Ideas side panel in PG Op. Import tab → Settings:Imports. Revisions → Weekly Detail Panel.
Read-only incomplete table. Two of six columns hardcoded zero.
Drop entirely. "Stuck orchestration" alert can surface in Settings:Admin or Workshop tracker section.
Conversational Insights
Concept stub. Hardcoded mock Q+A, no backend, never shipped.
Drop entirely. Verify with Sergey/Lee no stakeholder expectation.
Admin nav slot
Admin-gated menu entry. Tools fold into Settings.
Settings:Admin (hard-gated to cosmo-admin Cognito group).
Deferred to M3 / observability layer
27 observability jobs deferred to M3
M2 ships the operational restructure that makes the M3 observability split possible and coherent. Prerequisite: single canonical state machine + agreed metric definitions (AB-089). V1 has at least three incompatible status taxonomies and revision-count mismatch across four pages.
Workshop and Weekly are the two surfaces being designed. Both consume from a shared orchestration layer (Apollo events, build events, ideation events, state machine). Detail Panel mounts as a slide-over from either. Settings sits to the side as a small full-screen surface. Game Details panel is secondary. Dropped V1 fragments fold in lightly.
03Observability layer / M3
27 jobs deferred
M2 ships the operational restructure that makes the M3 observability split possible. Per-role wishlist below grounded in the vault. Implementation order decided when M3 planning begins, post-M2 ship.
PG
8 jobs
Quota pace per game (two-bar pattern from Status, M3-trustable)
Aging items: Ready-to-send sitting too long (P31, Einat 2-day threshold)
Revision rate per game and trend (Harel primary ask)
Partner-level delivery roll-up (revision rate by partner)
Customer
1 job
Receive structured delivery report: sent / approved / in revision (JTBD-15)
Yuval data question: contractual quota vs executed vs sent vs revised, per game / PG / total
Green-field
3 jobs
Translation chain quality from prompt to execution per game (Harel R2 thesis)
Quiet client churn detection before partner disengages (Fishi)
Expected delivery simulation given week plan + capacity (Einat)
Prerequisite before any M3 surface ships
Single canonical state machine + agreed metric definitions (AB-089). V1 has at least three incompatible status taxonomies and revision-count mismatch across four pages (34 / 104 / 181 / 244, P29). Observability numbers cannot be trusted until the data layer is reconciled.