Source: qdash /teams — c67 composition bake-off, 2026-08-14
The two coda stacks from v1, each rendered as a full composed page with the c67 decisions applied: Cost Center is canonical for spend (a slim team overview stays here and cross-filters the roster), the mapping workspace is the write path, and the credentials table is gone — relocated to /identity, leaving a per-person identity gist in the roster. Same Meridian Monday: felix's egress undecided, sofia's secrets.yaml finding unassigned, chen new this morning, $306 of $1,340 still unattributed.
The accountable roster
Risk-first console read: the roster is the page, everything else supports it. One merged caveat replaces the three honesty notes; the summary strip makes mapping coverage a headline number; teams appear twice — as a filter rail over the roster and as a slim spend overview that hands the deep dive to Cost Center.
| Person | Team | Risk story | Identity | Activity | Spend · 30d |
|---|---|---|---|---|---|
| fefelix@duck.com | Data | 1 Critical — undecided egress to metrics.coderelay.io · Sun | 2 agents · 31m ago | $70 | |
| sosofia | Platform | 1 Elevated — secrets.yaml read · Fri · unassigned | 2 agents · 2h ago | $98 | |
| prpriya | Payments | ok | 3 agents · 12m ago | $1692.4× median | |
| mamarcus | Platform | ok | 3 agents · 48m ago | $121 | |
| dadana | Data | ok | 3 agents · 1h ago | $112 | |
| chchennew today | ok | 1 agent · 19m ago | $4 | ||
| + 9 more, all ok — show → | $489 | ||||
Teams first
The eng-manager and FinOps read: teams are the accountable units, people nest inside. Cards lead, the roster arrives pre-filtered by a card click, and the mapping workspace opens as a drawer off the unassigned card. Adoption reads per team.
| Person | Risk story | Identity | Spend · 30d |
|---|---|---|---|
| fefelix@duck.com | 1 Critical — undecided egress · Sun | $70 | |
| dadana | ok | $112 | |
| sasam | ok | $52 | |
| totomas | ok | $35 |
Recommendation — Stack A, with B's card language
- J1 — "who is putting us at risk right now" — is the page's top job, and A answers it above the fold; B makes you click a card first.
- The IA doc classifies the page as a Fleet lens ("who owns the risk/spend?") — a roster question, not a rollup question.
- Only 4 teams exist today (free-text strings). Cards as the organizing unit flatter the data model more than it deserves pre-SCIM.
- B's best parts survive: the card language becomes A's slim team overview, and the promoted unassigned treatment becomes A's unmapped band.
- Mapping form factor — A shows a band, B shows a drawer; a sub-route is the third option. v3 renders all three on A's spine.
- Identity gist depth — one chip per row here; v3 tries 2–3 treatments (chip only · chip + method · summary line above the roster).
- Adoption strip earning its slot — fleet-level (A) vs per-team (B) vs cut.
- Cost Center EXP gate — the slim overview assumes /cost is reachable from the nav; flag at checkpoint.