Public report sample. Retained results from one dairy delivery, with local paths and source files omitted. Read the explanation.
Dairy plant - caller-mapped acceptance brief
Needs review · Next action: Review finding
Acceptance of the delivered OpenUSD scene against the owner brief, with explicit agent tolerances
2,081 prims, 1,105 meshes, 158 materials
137Passed check rows
0Failed check rows
7Unresolved declared requirements
10AI-routed human-review items
These categories overlap and use different units. Review items are not a count of defects or assets.
What needs attention
- Review the brief interpretation and its chosen targets or tolerances.
- 7 declared requirements remain unknown or in error; inspect their evidence gaps.
- AI triage routes 10 items to human review and 3 to more context.
What was considered
| Input or stage | Included status | Context |
|---|---|---|
| Brief provided | Yes | Dairy plant - caller-mapped acceptance brief · Requirements and sources |
| Brief provenance | Map authored by the producing agent from the owner-supplied brief; tolerances are agent decisions. | Recorded authorship; not independently authenticated. |
| Brief source files | 1 text | Text and reference images describe the requested result. |
| Brief interpretation review | pending | Producer-authored map; relevance and tolerances need owner review |
| Scene data | Saved USD inspected | dairy_plant.usda |
| Scripted checks | Completed | brief.measurements, materials, motion.connection, nvidia.asset-validator, openusd, scene.audit, textures.decode |
| Rendered views supplied | 0 recorded views | Supplied does not mean eligible for every review question. |
| Visual evidence and judge | Not requested | Caller-supplied views are separate from reference images in a brief. |
| Text context for triage | 2 selected text sources | See exactly what was considered |
| Producer decisions recorded | 0 | Only declared decisions in this assessment are counted. |
| Undeclared-assumption audit | No audit record attached | This report does not establish whether another audit ran. |
| AI use | Text triage only | Model opinions do not replace measurements or human decisions. |
Check results
| Basis | Checks | Pass | Warning | Fail | Unknown | Error | No subjects |
|---|---|---|---|---|---|---|---|
| General structural and delivery checks | 27 | 23 | 0 | 0 | 0 | 0 | 4 |
| Checks linked to human specifications | 114 | 114 | 0 | 0 | 0 | 0 | 0 |
Inspect every check, its subjects and measurements. Subject counts are kept per check and never summed into a unique-asset total.
AI risk triage
| Execution | Selected items | Low | Medium | High | Unrated | More context | Routine |
|---|---|---|---|---|---|---|---|
| Completed | 14 | 4 | 6 | 0 | 0 | 3 | 1 |
Low, medium and high prioritize human review; they are not verified engineering risk. Reasons, policy overrides and evidence.