perf(history): reduce redundant reads

[zerob13]

What changed

• Add an indexed hasMessages(sessionId) query using SELECT 1 ... LIMIT 1 and use it where session orchestration only needs a boolean.
• Treat missing file, link, and assistant-block buckets from successful bulk reads as authoritative empty results instead of issuing per-message fallback queries.
• Preserve legacy user-row/header fallback, assistant raw-content fallback, title generation, draft reuse, and conservative error handling.
• Keep the implementation rationale, boundaries, validation, and rollback notes in one document: docs/architecture/history-read-hot-path/spec.md.

Why

The send path was materializing full history for a boolean check. Structured history assembly also repeated single-message queries after successful bulk reads returned empty buckets. The earlier isolated baseline measured the smallest fixture moving from 8 full reads / 120 SQL statements to 7 / 105 after the existence query, then to 7 / 35 after removing the empty-bucket N+1 fallback. This PR ports only those two low-risk fixes onto the latest dev; it does not include the benchmark platform or broader audit work.

Impact

• Less synchronous SQLite work before sending a message.
• Query counts no longer grow per message solely because file/link/assistant bulk buckets are empty.
• No database migration, cache, runtime flag, event bus, observer, history projection, or snapshot lifecycle change.

Validation

• Targeted Vitest: 307 passed, 2 native-ABI skips.
• Electron native SQLite: 3/3 passed; the existence query uses the existing session index.
• pnpm run typecheck:node
• pnpm run format:check
• pnpm run i18n
• pnpm run lint
• git diff --check origin/dev...HEAD

Full suite, E2E, and full build were intentionally not run; the change is covered by focused contract tests and real SQLite verification.

Summary by CodeRabbit

• Performance Improvements

  • Session history checks now determine whether messages exist without loading the full message history.
  • Reduced unnecessary reads when preparing messages for sending.
  • Missing structured message data is handled as empty content without extra lookups.

• Tests

  • Added coverage for message-existence checks, indexed queries, and structured-content loading behavior.

• Documentation

  • Added architecture documentation describing the history-read optimization and verification criteria.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: e2ba4b94-87f3-4493-98ad-8fe546de8f3c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• ✅ Review completed - (🔄 Check again to review again)

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/history-read-hot-path

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}