Detect likely leaked API keys, tokens, private-key headers, JWTs, and credential assignments in caller-supplied source or configuration text. Return type, line, fingerprint, and a fully redacted preview with no network access or code execution.