Summary
Add an annotation-quality audit mode and local CI tool that tells developers whether their Askable context is useful, safe, stable, and within budget.
Proposed UX
<AskableInspector audit={{
sensitiveKeys: ['email', 'token', 'secret'],
maxElementTokens: 200,
requireStableIds: true,
}} />
npx askable audit http://localhost:3000 --journey ./askable.journey.ts --format json --fail-on error
Diagnostics
- malformed metadata
- unstable or non-unique selectors/IDs
- suspicious sensitive keys and unredacted packet fields
- empty or flattened/noisy text
- excessive token cost
- missing accessible labels
- important interactive regions without annotations
- chart/table regions that should use curated
data-askable-text
Acceptance criteria
Summary
Add an annotation-quality audit mode and local CI tool that tells developers whether their Askable context is useful, safe, stable, and within budget.
Proposed UX
Diagnostics
data-askable-textAcceptance criteria