docs(evaluation) 15/15: feature guide and a sample dataset

[Ahmath-Gadji]

Part 15 of 15 of the split of #811. Targets eval/14-ui-page (#825). Docs and data only — no code.

What

• CLAUDE.md — the Evaluation section: the flow, the layering rule (the service dispatches through a port and never touches Ray), why there are two promptfoo configs, and which values are deliberately not configuration.
• tests/evaluation/ — a ready-made test set: 11 questions over 6 documents, every answer checked against the source rather than generated.

Notable

• The corpus is not committed — third-party PDFs, ~36 MB. corpus.txt lists the filenames and the README gives the two commands to assemble and upload it.
• 24 documents for 11 questions, on purpose. Only 6 are the subject of a question; the other 18 are distractors, each topically adjacent to a question's source (other AI-policy papers, other gut/neuro medical papers, other agricultural releases). A corpus where every document is on a different subject makes retrieval look better than it is — any half-working retriever scores a perfect hit rate when there is one candidate per topic.
• The repo ignores *.csv, so the committed test set is un-ignored explicitly. It is source, not scratch data.

Carried over from #811 — known limitations, unchanged

Flagged for a decision rather than fixed:

• Indexing is sequential, so files_per_minute measures serialised throughput, not the system's real concurrency.
• /search defaults similarity_threshold to 0.75 and the eval does not override it, so retrieval can return fewer than top_k and depress recall for reasons unrelated to ranking.
• A separate Ray cluster needs <data_dir> on shared storage — the runner reads dataset files by path.
• Ordinary users can still create __eval_-prefixed partitions, which are merely hidden from listings.

Testing

ruff and the layer-import guard pass. No code changed.

[coderabbitai]

Warning

Review limit reached

@EnjoyBacon7, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 59 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: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 5272a288-5c96-4402-a5b7-bfa386efc6ad

📥 Commits

Reviewing files that changed from the base of the PR and between 7f1f006 and 27e5220.

⛔ Files ignored due to path filters (1)

• tests/evaluation/rag_dataset_sample.csv is excluded by !**/*.csv

📒 Files selected for processing (4)

• .gitignore
• CLAUDE.md
• tests/evaluation/README.md
• tests/evaluation/corpus.txt

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch eval/15-docs-and-sample-dataset

---

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.}

[Ahmath-Gadji]

@coderabbitai full review

[coderabbitai]

✅ Action performed

Full review finished.

---

Your included review limit is currently reached under our Fair Usage Limits Policy. This review may still proceed through usage-based billing if eligible. Your next included review will be available in 56 minutes.