Fix Messages API cache breakpoint placement

[chrarnoldus]

Summary

• place Messages API cache breakpoints on the final cacheable content block
• remove ineffective top-level cache control while preserving its TTL
• document that Vercel AI Gateway does not honor top-level cache control

Verification

• Finished in 18ms on 2 files using 2 threads.
• focused Jest attempted but blocked by unavailable PostgreSQL during global setup
• web typecheck attempted but exceeded the 120-second environment limit

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

Reviewed the refactor moving Messages API cache breakpoints from the ineffective top-level cache_control onto the last cacheable content block, with logic and edge cases (thinking-only blocks, string content, preserved TTL) covered by matching test updates; no high-confidence bugs, security issues, or breaking changes found.

Files Reviewed (2 files)

• apps/web/src/lib/ai-gateway/providers/openrouter/request-helpers.ts
• apps/web/src/tests/openrouter-request-helpers.test.ts

---

_{Reviewed by claude-sonnet-5 · Input: 14 · Output: 4.9K · Cached: 258.8K}

_{Review guidance: REVIEW.md from base branch main}