Loom's own audit of two codebases by the same team
BeforeLoom's control areabuilt before we had a standard
AfterLoom Basewritten with Loom's method, open source
Overall score2275Holding together
- Foundations and tokens
4988Holding together - Components
896Tightly woven - Patterns and pagesnot assessed60Taking shape
- Usage
1097Tightly woven - Governance
2676Holding together - Brand, illustration, microcopy and localisation
4226Loose threadslower - Tools and sync
554Loose threads - AI readiness
1581Holding together
Top finding in the area
- Components · After69 components, one contract each. 453 of 453 parity tests pass.
- Usage · After3 UI elements used in the consumer with no system contract or rule
- AI readiness · AfterAgent retrieval is evaluated once, not tested on every change
For reference: takenote, an open-source notes app with no design system, scores 16.
Real audit output from Loom's own open rubric, low scores included: two codebases by the same team, not one system before and after. We are moving the control area onto Loom Base now. Each score links to its checks and the command that produced it.