feat(retrieval): one search path, recency decay, honest degradation

[plind-junior]

step 3 of the roadmap (.superpowers/roadmap.md): make good retrieval the default and stop lying about degradation.

• kb.search had three copies of the backend waterfall (mcp, jsonl, cli) and they had already drifted — fusion landed in kb.context while kb.search auto still waterfalled embedding-first. context.search_kb is now the single implementation; the mcp tool and jsonl handler delegate to it. auto fuses embedding + fts5 via rrf and falls back to substring only when both are empty; an omitted backend defers to retrieval.backend in config.yaml (the starter default is hybrid).
• honesty block: search and context responses carry retrieval = {configured, used, semantic_available, degraded}. a base install (no embeddings extra) serving lexical hits reports used=fts5 + degraded=true instead of a hollow hybrid label.
• retrieval.recency: a config-gated half-life decay (default 90 days, whole-day quantized so same-day scores are deterministic) blended into fused context-pack scores — fresher knowledge outranks equally-relevant stale knowledge. rescoring-only; an artifact never vanishes from the pack. enabled in the starter config for new kbs; existing kbs keep byte-identical ordering until they add the key.

covered by new tests in test_retrieval_backend.py (recency on/off, degradation reporting, config-default backend, substring fallback, mcp/jsonl parity through search_kb). full local gate green: pytest, mypy, ruff.

[coderabbitai]

Warning

Review limit reached

@plind-junior, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 33 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: a1daef7c-0e0f-4db2-8cc5-393e28e4ea43

📥 Commits

Reviewing files that changed from the base of the PR and between f1826ac and e7d6e76.

📒 Files selected for processing (6)

• src/vouch/context.py
• src/vouch/index_db.py
• src/vouch/jsonl_server.py
• src/vouch/server.py
• src/vouch/storage.py
• tests/test_retrieval_backend.py

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/retrieval-defaults

---

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.}