Context. surface studio · domain isis · route /studio/isis/benchmarking · auth signed-in + studio entitlement (segment not in AAA_ONLY_STUDIO_ROUTES — renders as-is) · source apps/oshun/web/src/app/studio/isis/benchmarking/page.tsx
Last walked. 2026-05-29 automated runtime walk (Playwright headless) — render, /v1 data (2xx), console/page-errors, expected content, screenshot verified; live screen-reader, touch, offline, and telemetry-delivery checks pending a manual AT pass. Evidence: WALKTHROUGH/results/runtime-sweep-2026-05-29.md; body re-derived 2026-06-03 from current source (lane-console architecture)
Purpose#
Admin lane console for §24.9 quality benchmarking: it scores generated 3D outputs
against reference embeddings using the real 3d-generation-benchmarks
CLIP-similarity metric (clamped cosine) and gates each case on a per-case
threshold. All reads and evaluations flow through /v1/admin/isis/benchmarking;
it is admin-scoped and fails closed.
Entry points#
- Sibling quick-action from
/studio/isis/artifact-detection - Studio index (
/studio) - Direct URL / bookmark
Layout regions#
page.tsx mounts <ShellLayout active="studio"> (no breadcrumb panel) and three
panels: the workspace, the Route Map, and sibling quick-actions.
- Header: shell header
- Workspace panel (
data-isis-benchmarking-workspace):<h1>(WorkspaceHeading) "Isis Benchmarking Workspace"- Summary paragraph (
data-benchmark-summary) - Rollup pass-rate (
data-benchmark-rollup/data-benchmark-passrate/data-benchmark-passing) - "Evaluate a benchmark case" form (
data-benchmark-form) - Stored cases list (
data-benchmark-cases), eachdata-benchmark-case; empty statedata-benchmark-cases-empty
- Route Map panel (
data-isis-benchmarking-route-map):<h2>Route Map</h2>overSTUDIO_ISIS_BENCHMARKING_ROUTE_MAP(5 entries) - Sibling quick-actions panel:
/studio/isis/artifact-detection,/studio
States#
- Loading —
data-benchmark-loading - Unauthorized — 401/403 admin-scope fail-closed;
data-benchmark-unauthorized - Error — non-OK or malformed response;
data-benchmark-error - Ready (form) — response present; rollup + form + cases render
- Result — after a score POST,
data-benchmark-resultshowsdata-benchmark-verdict,data-benchmark-clip(CLIP similarity), anddata-benchmark-status
Interactions#
Evaluate a benchmark case (data-benchmark-form)#
- Case id (
data-benchmark-caseid, text input) - Label (
data-benchmark-label, text input) - Threshold (0..1) (
data-benchmark-threshold, number input,min=0 max=1 step=0.01) - Reference embedding (comma-separated numbers)
(
data-benchmark-reference, text input) - Candidate embedding (comma-separated numbers)
(
data-benchmark-candidate, text input) - Run benchmark (
data-benchmark-run, button; "Scoring…" while submitting) —POST /v1/admin/isis/benchmarking
Route Map#
- Route-map articles ×5
Sibling quick-actions#
- Open Isis Artifact Detection workspace →
/studio/isis/artifact-detection - Back to Studio workspace index →
/studio
Data & contracts#
- Reads:
GET /v1/admin/isis/benchmarking(cases + rollup pass rate) - Writes:
POST /v1/admin/isis/benchmarking(score a case; returns the CLIP similarity verdict and updated rollup) - Realtime: none
- Caching: client fetch on mount,
cache: 'no-store',buildBffAuthHeaders() - Auth/role check: admin-scoped, fail-closed (401/403); route gated on signed-in + studio entitlement
- Component sources:
apps/oshun/web/src/components/studio/StudioIsisBenchmarkingWorkspace.tsxapps/oshun/web/src/components/studio/StudioIsisBenchmarkingRouteMap.ts
Cross-references#
- Studio overview:
../../studio-overview.md - Sibling routes:
Open questions / known gaps#
- No breadcrumb on this route — navigation is via sibling quick-actions
- Embeddings are pasted as raw comma-separated vectors; confirm whether they will be sourced from real asset embeddings before V1