Recipe 30: Citation verification #30

Recipe 30 of 60 · Level 3 (Intermediate) · Search & retrieval · Choice

Use case

Check each answer claim against its cited passage and flag unsupported or contradicted claims while Python verifies citation identifiers.

What to build

The folder recipes/30-citation-verification/ with notebook.ipynb, README.md, and fixtures/, meeting the recipe contract. Start from python tools/new_recipe.py 30 and the template in recipes/_template/.

Build notes

• Python first verifies that every citation identifier exists and resolves. That step never reaches the model.
• One Choice per claim against its cited passage: supported, unsupported, contradicted.
• Include a correct claim that cites the wrong passage.

Decision design

• Choice. Options are a fixed set supplied by Python, including the fallback outcome this use case names. Show the choice, the per-option probabilities, and the confidence for individual examples before aggregating.

Level 3 scope

Multiple judgments or stages, structured evidence, deterministic composition, and task-level evaluation.

Acceptance criteria

• recipes/30-citation-verification/notebook.ipynb runs top to bottom offline, with no key and no network, and its outputs are committed
• Live mode is opt-in (JEV_COOKBOOK_LIVE=1 with TYPESAFE_API_KEY) and uses the same question definitions as the offline run
• Fixtures are synthetic, include the hard cases named in the build notes, carry gold labels, and pass the fixture validator
• Every response fixture records its provenance, and the notebook states which mode it ran in
• Evaluation: Claim-level F1 for the unsupported and contradicted classes.
• The deterministic composition lives in a Python function with tests in tests/
• Evaluation is at task level, not only per question
• No statement about Jev's quality, latency, or cost appears without recorded live inference; in a synthetic run, metrics are labelled as pipeline checks
• recipes/30-citation-verification/README.md says what the recipe teaches, how to run it, and what was and was not measured live
• The pull request changes only recipes/30-citation-verification/ and CI is green

Dependencies

• Blocked by Foundation: CI executes every notebook offline and validates fixtures #69 (CI executes every notebook offline and validates fixtures)
• Builds on Recipe 20: Claim support classification #20 (Claim support classification). Merge those first and reuse their patterns; do not copy their fixtures.
• Tracked in Epic: Level 3 · Intermediate recipes #77

Sources

Read these before writing the questions.

• S02 · TypeSafe AI: Primitives (Questions): Choice, Noul, Score, independent questions, and dependent follow-up calls.
• S03 · TypeSafe AI: Confidence: Distribution-derived confidence and task-specific thresholds informed by observed outcomes.
• S04 · TypeSafe AI: How to build with TypeSafe: Narrow semantic judgments integrated with workflows, deterministic logic, and side effects controlled by code.
• S06 · TypeSafe AI: Cookbooks: Worked patterns for extraction, citation checking, skill selection, hierarchical classification, and feature discovery.

Out of scope

• Changes to shared code, the catalog, or the root README. If the shared package is missing something, open a foundation issue and link it here.
• Real side effects of any kind. Actions are simulated.