Semantic Drift Detector MCP

Choose this endpoint when you need to programmatically detect whether text, AI outputs, or documents have shifted in meaning, tone, or conceptual focus over time or revisions. It is particularly useful for AI monitoring pipelines, content quality assurance workflows, and catching when language models begin producing semantically inconsistent outputs. Prefer it over general-purpose text analysis when the specific question is about drift or deviation from a prior meaning baseline.

• Empty or null 'data' field returns validation error
• Text too long may exceed processing limits
• Ambiguous or very short input may produce low-confidence drift scores
• No baseline reference may limit comparative drift analysis
• Non-textual binary data passed as string will produce meaningless results