Tags: ranuts/document

chore(seo/geo): allow AI training crawlers in robots.txt

Switch from 'citable-not-trained' to allowing all crawlers, including AI
training crawlers (GPTBot, ClaudeBot, Google-Extended, CCBot, …). Rationale:
being in the training corpus lets assistants recommend the editor from memory,
not only via live retrieval. Privacy is unaffected — this governs the public
landing pages only; user documents are never uploaded.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

test: add SW routing tests covering font-interception crash prevention

Documents the fetch routing rules in public/sw.js that prevent the
OnlyOffice v7.5 units_per_EM crash:
- Font files (.ttf/.woff/.woff2/.otf/.eot) must bypass the SW to avoid
  added latency that triggers Chrome's font-loading intervention
- Document URLs (?src=, ?file=) bypass the cache to stay fresh
- Cross-origin and non-GET requests are not intercepted

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

chore: update CI workflow to run TypeScript linting instead of genera…

…l linting

chore: clean up whitespace in UI component to improve code readability

fix: add null check for performance.memory.usedJSHeapSize before send…

…ing analytics data in document, presentation, and spreadsheet editors