[cixzhang]

Summary

• defines five observable component-style authoring invariants: StyleX lowering, CSS-over-JS presentation ownership, valid static composability, scoped relational state, and named wrapper responsibility
• keeps exact StyleX capabilities, shorthand/reset/null syntax, helper names, and lint matchers as versioned guidance/tooling rather than permanent architecture vocabulary
• permits component-owned non-glyph inline SVG while delegating semantic glyphs to icon authority
• permits wrappers for semantics, behavior/ref, formatting contexts, outside-border spacing, or a required style-bearing DOM boundary when the child has no compatible surface
• delegates tokens, theming axes/targets, BaseProps/public styling, RTL, and RSC module boundaries to their canonical owners

Audit-row coverage

Owned here:

• Q6 / T14 — wrapper responsibility
• durable composability outcome behind Q7
• T8 — StyleX lowering and CSS-over-JS ownership
• valid-lowering outcome behind T9
• T10 — CSS-over-JS ownership
• T11 — relational-state isolation

Delegated:

• Q4 / T17 / T18 → icon architecture, Icon contract, and theme authoring
• T2b → theme-token architecture
• T13 → future current bidirectionality owner
• T19 / T27 → component-theming-surface; T19 reachability also uses public-component-api
• T20 → public-component-api
• T30 → future current RSC/module-boundary owner

Owner assumptions

• Inline SVG is prohibited only when it shadows a registry-owned glyph; component-owned non-glyph vectors remain allowed.
• A wrapper needs a named runtime responsibility; source clarity alone is not normative.

Overlap receipt

• Searched current records and open PRs by owner/id, StyleX, CSS, wrapper, marker, raw CSS, variant axes, and affected paths.
• feat(theme): ordered environmental adaptations in defineTheme #5543, Validate theme visual props from generated Core contract #6061, and Spec: generate the visual-prop contract from TypeScript #6091 overlap theming/visual-prop authority only; this record delegates those rows and does not modify their files.
• Icon scope overlaps current icon architecture and open fix(richtext): require a RichTextView label and split read-only from disabled (#5160, stack 1/6) #5204; no icon authority is changed here.
• docs: define React component runtime architecture #6122 is adjacent React runtime authority but explicitly delegates CSS presentation to this owner.
• docs: define RTL audit evidence policy #6046, docs: define bidirectionality design intent #6047, and docs(core): define the internationalization contract #6048 remain untouched.

False-positive / false-negative audit

Initial bounded current-code matrix: 12 correct, 1 false-positive risk, 1 false-negative risk, 2 untestable.

After exact fixes for child style-bearing surfaces, component-local raw CSS, and StyleX registry/package version drift: 11 correct, 0 false positives, 0 false negatives, 5 intentionally untestable capability-dependent cases across the same 16 cases. Untestable cases require matching the capability registry's recorded StyleX version to the installed package.

Validation

• node scripts/check-knowledge.mjs
• pnpm check:sync
• Prettier
• git diff --check
• independent authority map, Matt-style challenge, and exact-file current-code FP/FN audit

No implementation, tooling, lint, wiki, or audit-table changes. No changeset.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Sep 8, 2026 12:27am UTC

[github-actions]

PR Analysis Report

📚 Storybook Preview

View Storybook for this PR
GitHub Pages may take up to a minute to hydrate after deploy.

🧪 Sandbox Preview

View Sandbox for this PR
GitHub Pages may take up to a minute to hydrate after deploy.

No new or modified components detected.

Bundle Size Summary

No component packages changed.

Accessibility Audit

Status: No accessibility violations detected.

---

_{Generated by PR Enrichment workflow | Storybook | Sandbox | View full report}

[cixzhang]

/approve-spec 55d4568

[cixzhang]

Semantic verdict: request changes

Thanks — the authority boundaries are clear. I think the verification contract needs one more pass before this becomes current. The relational scanner only compiles through the global default marker, so it never proves nested-owner isolation. The wrapper rule also recommends xstyle for any relative import without proving the child has a compatible style-bearing surface.

Could we either add the scoped-marker/nested-owner and incompatible-child fixtures the Verification table names, or label those rows as current verification gaps?

[Reviewed by Robohands]