Source and example scope
Compare each saved result with its authored expectation. A disagreement remains in the fixture review; the call is not repeated to obtain a preferred label.
Provide the prior resolution, its conditions and the current question. Compare a reusable answer with changed conditions or a genuinely different decision.
The first two presets state both questions, the recorded resolution, its required condition and the current facts. Jev compares this supplied record. It does not retrieve a decision log or decide whether the original resolution was correct.
A decision log records why a choice was made. Reusing the label without that condition can carry an outdated conclusion into a new task. The XML example exposes a changed condition rather than treating both tasks as identical exports.
Different question keeps unrelated decisions apart even when they share a word such as archive. Unclear applicability leaves an unresolved condition visible instead of assuming that nothing changed.
The output is context for a review step. It does not replace mandatory rules, silently approve a current action or prove that the recorded resolution was correct in the first place.
These fictional inputs and decision rules were written for this page. The linked X discussion provides background; the technical implementation follows TypeSafe's Choice contract.

Use the sample to understand the rule, then test your own input for decision reuse.

Provide the prior resolution, its conditions and the current question.

Open the decision rules and keep the intended distinction between Reusable resolution, Changed conditions, Different question, Unclear applicability, Need evidence. Edit the choices only when testing your own rule.

Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

Review Unclear applicability, Need evidence before using a result. The label is a decision about the supplied information, not an executed action.
| Original input | Observed label | Leading probability |
|---|---|---|
| Same approved condition | Reusable resolution | 94.0% |
| Target changed | Changed conditions | 100% |
| Different decision | Different question | 96.0% |
| Condition unresolved | Changed conditions | 78.0% |
| Same words, different purpose | Different question | 96.0% |
| No prior resolution | Need evidence | 100% |
These are six original synthetic inputs, tested once per request revision. The percentage is the leading option probability, not measured accuracy. Authored expectations and any disagreements remain in the downloadable evidence. 5 of 6 results matched the authored fixture expectations. For Condition unresolved, the expectation was Unclear applicability; the observed choice was Changed conditions.Verified
Compare each saved result with its authored expectation. A disagreement remains in the fixture review; the call is not repeated to obtain a preferred label.
The single-request SDK runner uses your private TYPESAFE_API_KEY. The offline consumer reads saved observations, checks request hashes and prints a pending output. It does not call the model or change any external system.
npm install @typesafe-ai/sdk@0.6.0
# Set TYPESAFE_API_KEY privately in your shell.
node run-choice-sample.mjs samples.json case-1
python preview_choice_workflow.py samples.json saved-api-results.jsonThe rubric explains the decision; the application owns the next step.
Freshness asks whether a note still reflects current evidence. This question asks whether a recorded resolution answers this particular current decision under its stated conditions.
No. It supplies relevant prior context. The current application still owns approvals, mandatory rules and execution.
No. It is the model distribution for this one request. The six examples are functional checks with authored expectations, not a representative benchmark. Inspect disagreements and evaluate your own examples before using the label.
The window shows an error and supplies no fresh judgment. The downloadable runner sends one request without retries. Need evidence is a valid outcome about incomplete input; it does not stand in for a network or provider failure.
Edit the evidence above or use the Playground to test another bounded question.