Back to Prompt Library
evaluation
Review evidence and failure behavior
Inspect the original prompt language first, then copy or adapt it once you know how it fits your workflow.
Linked challenge: Detect an Unsupported Claim in an Exhibit Label
Format
Text-first
Lines
4
Sections
1
Linked challenge
Detect an Unsupported Claim in an Exhibit Label
Prompt source
Original prompt text with formatting preserved for inspection.
4 lines
1 sections
No variables
0 checklist items
Review the implementation against this contract: A date range does not support a single exact year. Keep attribution qualifications from the approved record. Test this domain failure: A web page repeats the stronger claim without primary evidence. Do not upgrade certainty. Check input citations, calculations, and actual tool logs. Report unsupported claims and unexecuted tests. Do not approve external actions.
Adaptation plan
Keep the source stable, then change the prompt in a predictable order so the next run is easier to evaluate.
Keep stable
Preserve the source structure until you know which part of the prompt is actually driving the result quality.
Tune next
Change domain facts, examples, and tool context first before you rewrite the instruction scaffold.
Verify after
Validate one failure mode at a time so prompt changes stay attributable instead of getting noisy.