Case Credibility Score — Turkish SMB AI Cases

An evaluation skill that analyzes artificial intelligence success stories (case studies) aimed at SMBs in Turkey across three axes and converts them into a credibility score between 0 and 1. The axes are: (1) primary-source strength — how many independent sources corroborate the claim, (2) metric verifiability — whether the cited numbers come with a baseline, measurement window, and methodology, (3) sales-magnet (hype) pattern present/absent — detection of marketing tricks such as inflated round numbers, unquantified praise, and false urgency.

When is it used? To quickly tell whether the "we delivered an X% improvement at this company" type of case shared by an AI vendor or competitor is genuine or a marketing claim; to close the evidence gaps in cases you are about to publish before they go live; to filter out weakly-evidenced examples when building a case library.

Output: A Markdown score report — a weighted score card by axis (table), primary-source/metric/pattern analyses, a reliability-band verdict (reliable / conditional / suspicious), and concrete remediation suggestions to raise the score. It applies heuristic thresholds such as a ceiling rule for single-source cases; the final score is normalized to the 0-1 range.