Zentric Protocol Mcp

Adds prompt injection detection and PII scanning to Claude with sub-millisecond response times. Exposes IntegrityGuard for catching jailbreaks across 22 attack signatures and PrivacyGuard for identifying 12 PII types with regional format validation (SSN, IBAN, NIF, CPF). Every request returns a signed ZentricReport with verdict, matched patterns, and anonymized output. Useful when you're building agent pipelines that handle user input and need deterministic security checks before LLM execution, or when you need GDPR Article 30 audit trails without adding latency. Requires a Zentric API key. The repo contains middleware and schemas but the detection engine runs on their infrastructure.