[Raing5Days]

Summary

Performance optimizations discovered and verified via autonomous flywheel scanning, targeting high-frequency string splitting, regex parsing, and YAML document caching in Web UI.

Key improvements:

1. Global Search Matching Query Tokenization (searchMatching.ts):

   • Hoisted queryTokens invocation outside the candidate fields iteration loop (previously splitting the query up to 5 times per item across hundreds of items).
   • Added bounded LRU cache for queryTokens and fieldWordsCache for term/acronym word splits, eliminating repeated regex splitting and array allocations during live search keystrokes.

2. Scheduled Jobs DateTime Parsing & Formatting (localizedDateTime.ts):

   • Reused hoisted static regex patterns FORMAT_PATTERN and DIGIT_PATTERN instead of re-instantiating regexes on every call.
   • Added bounded format and parse memoization cache, enabling instant rendering for scheduled jobs tables and timeline rows with zero regex recalculations.

3. My Agent Identity Document Memoization (identityDocument.ts):

   • Added parsedDocCache memoizing parseIdentityDocument by content string, returning identical object references (===) across repeat renders and property accesses without repeated markdown/YAML parsing.

Type and Areas

Performance / refactoring. Web UI (src/web-ui).

Verification

• node scripts/run-vitest.mjs run src/app/global-search/searchMatching.test.ts -- 6 passed (100%).
• node scripts/run-vitest.mjs run src/app/components/scheduled-jobs/localizedDateTime.test.ts -- 12 passed (100%).
• node scripts/run-vitest.mjs run src/app/scenes/my-agent/identityDocument.test.ts src/app/scenes/my-agent/useAgentIdentityDocument.test.tsx -- 6 passed (100%).
• node scripts/check-repo-hygiene.mjs -- passed (clean repository hygiene and formatting).

Checklist

• Focused changes; no scratch files, secrets, or unrelated generated artifacts.
• Preserves pure functional contracts and existing testing guarantees.