What the instrument measured.
8 / 2 legibility
3 / 5 constraint
8 / 9 interaction
Measured against a pinned commit.
- Repository
- microsoft/fluentui
- Revision
- f51ba3576887, committed 2026-09-08
- System directory
- packages/react-components
- Components found
- 527, of which 261 carry documentation and 215 have a colocated doc artifact
- Agent context
- 1 context file, 13 skills
- Tokens
- 661 distinct, across 44 sources (44 authored); dark mode generated
- Adherence surface
- 6,662 files outside the system, graded own-docs-site: apps/ (docs, Storybook, perf tests).
What the counts point at.
Each finding carries its count, its severity and up to three file references. A finding on a property that was not banded is still a finding; it is reported without a score.
The design system's own component source hardcodes colour and size values instead of referencing its own tokens. This happens in 2139 of 2149 total colour/size references in system source (99.5%), so an agent copying that file as a pattern copies the hardcoding along with it.
- packages/react-components/react-migration-v8-v9/library/src/components/Theme/themeDuplicates.ts:158 #000000
- packages/react-components/react-skeleton/library/src/components/SkeletonItem/useSkeletonItemStyles.styles.ts:102 4px
- packages/react-components/react-avatar/library/src/components/Avatar/useAvatarStyles.styles.ts:33 32px
Design tokens are declared in more than one place in this repository's own authored source, so a value changed in one must be kept in agreement with the others by hand.
- packages/react-components/react-theme-sass/sass/colorTokens.scss sass, 184 token(s)
- packages/react-components/react-theme-sass/sass/colorPaletteTokens.scss sass, 182 token(s)
- packages/react-components/react-headless-components-preview/stories/.storybook/tokens.css css, 71 token(s)
Some components carry no documentation an agent could find. That is 266 of the 527 components found in system source, with no head comment on their own source, no colocated doc artifact matching their name, and no docs/ entry, so an agent must reverse-engineer their contract from implementation alone.
- packages/react-components/theme-designer/src/components/ColorTokens/AccessibilityList.tsx
- packages/react-components/theme-designer/src/components/Sidebar/AccessibilityPanel.tsx
- packages/react-components/react-accordion/library/src/AccordionHeader.ts
The application hardcodes colour values instead of using the design system's own tokens, so an agent editing nearby code has no signal that a token exists and will likely repeat the pattern.
- apps/public-docsite-v9/src/shims/ThemeShim/themeDuplicates.ts:157 #000000
- packages/tokens/src/global/colors.ts:7 #050505
- apps/public-docsite/src/pages/Styles/Colors/palettes/sharePointThemes.ts:10 #004578
The application defines its own component under the same name as one the design system already ships. This happens for 85 of the 952 component names the design system exports, so an agent importing that name by convention can silently get the wrong implementation.
- apps/public-docsite/src/components/Nav/Nav.tsx:18 Nav
- apps/public-docsite/src/components/Table/Table.tsx:2 Table
- apps/theming-designer/src/components/Header.tsx:95 Header
Some screens render a raw HTML element in place of the system component built for it. This happens in 230 of 249 total uses of a replaceable element (92.4%), so its styling and behaviour can drift independently of the system.
- packages/react-examples/src/react-charting/VerticalStackedBarChart/VerticalStackedBarChart.DateAxis.Example.tsx:146 <input>
- packages/react-examples/src/react-charting/VerticalStackedBarChart/VerticalStackedBarChart.Styled.Example.tsx:121 <input>
- packages/react-examples/src/react-charting/GroupedVerticalBarChart/GroupedVerticalBarChart.Truncated.Example.tsx:127 <input>
Built twice: cold, then informed.
Set up from its own documentation, in full, with every step that could not be followed recorded. Draw 1 of 3, judged; draws 2 and 3 pending.




What setting it up found
The README's documented setup uses teamsLightTheme and does not mention dark theming, theme switching, or that a dark counterpart (teamsDarkTheme) exists in the same package.
Handling: Followed literally: teamsLightTheme only, exactly as documented. Not silently upgraded. If a generated screen mis-renders in dark mode, that is attributable to Fluent's documented setup rather than to the agent.
The package README documents installation and provider wiring only. Component APIs, layout primitives and composition guidance live on an external documentation site, not in the installed package.
Handling: No action. Recorded because the informed arm points the agent at the system's documentation, so what the package itself carries versus what requires leaving it is exactly what the two-arm design measures.
Screens generated
| Arm | Draw | Date | Duration | Files read | Outcome |
|---|---|---|---|---|---|
| cold | 1 | 2026-08-14 | 19 min | 10 | completed |
| informed | 1 | 2026-08-14 | 10 min | 2 | completed |
Model: claude-opus-4-8. Generated under GitHub Actions isolation, the agent's own environment, with the frozen task text and no interaction. Files read is the number of repository files the agent opened, from its own transcript.
Cold arm, draw 1 · the working
- Counted
- 0 hardcoded colours, 8 hardcoded dimensions, 0 raw elements where a component exists, 76 uses of the system's components, 0 foreign UI imports, 0 invented names.
- Read
- 0 documentation reads, 11 reads of the package's source or types; went looking for documentation unprompted: no.
- Checklist
- I1 yes · I2 yes · I3 yes · I4 yes · I5 yes · I6 yes · I7 yes · I8 yes · I9 yes · I10 noI10: every control keyboard reachable, not observed (probe tab walk: 13 toggles, 2 time inputs, 1 selects reached, save not reached, focus visible on 100%)
- Composition
- Six no's: conforms. Cards on the neutral ground, inset, gaps. The Switch thumbs render as filled pills without a visible knob at 1280 (a theming or icon-font artefact, not a layout one).
Informed arm, draw 1 · the working
- Counted
- 0 hardcoded colours, 15 hardcoded dimensions, 1 raw elements where a component exists, 94 uses of the system's components, 0 foreign UI imports, 0 invented names.
- Read
- 0 documentation reads, 14 reads of the package's source or types.
- Checklist
- I1 yes · I2 yes · I3 yes · I4 yes · I5 no · I6 yes · I7 yes · I8 yes · I9 yes · I10 noI5: unsaved changes indicated, not observed (probe: changed a control; unsaved wording did not appear). I10: every control keyboard reachable, not observed (probe tab walk: 13 toggles, 2 time inputs, 1 selects reached, save not reached, focus visible on 100%)
- Composition
- Six no's: conforms. Cards with icons, a floating save bar, inset, gaps. Clean.
Judged 2026-09-09 by an agent under authorisation; a second read is pending. Constraint from the counts, banded twice, on how many kinds of the system's rules the output broke and on how many times it broke them, with the worse of the two published (deviation of 2026-09-10). Legibility and completeness from the transcript and the source, interaction from the ten-item checklist, composition from the six contract questions. Hover a checklist item for its observation.
Clone the repository at the revision above, point the instrument at the system directory, and the same numbers come out. The field manifest, the run script and the full payload this page was built from are committed beside it. Methods.