Recipe 15: Sensitive text triage #15

Recipe 15 of 60 · Level 2 (Easy) · Trust & security · Noul

Use case

Flag synthetic documents that may contain personal information so a simulated review queue can prioritize redaction checks.

What to build

The folder recipes/15-sensitive-text-triage/ with notebook.ipynb, README.md, and fixtures/, meeting the recipe contract. Start from python tools/new_recipe.py 15 and the template in recipes/_template/.

Build notes

• Synthetic documents only. Every name, number, and address is fabricated and obviously so.
• One proposition: the document may contain personal information.
• Python orders a simulated review queue by the noul value.
• Say plainly that this prioritises review and does not guarantee detection or perform redaction.

Decision design

• Noul. Write each proposition as a statement that is true or false. One independent question per label or concern, sent together in one request. Noul has no confidence field, so each threshold is chosen from examples and evaluated.

Level 2 scope

Context-sensitive matching, scoring, or verification with candidate sets and explicit uncertain or no-match outcomes.

Acceptance criteria

• recipes/15-sensitive-text-triage/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: Recall of sensitive documents within a fixed review budget, and precision at the chosen threshold.
• Candidate sets are assembled by Python and shown to the reader
• The uncertain or no-match outcome is exercised by several fixtures and reported separately in the evaluation
• 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/15-sensitive-text-triage/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/15-sensitive-text-triage/ 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)
• Tracked in Epic: Level 2 · Easy recipes #76

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.
• S07 · TypeSafe AI: Jev 1.13 jaggedness: Version-specific limitations affecting arithmetic, context selection, adversarial content, and generation.

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.