Shield FAQs

Question 1

Is Shield compatible with specific LLM clients or protocols?

Accepted Answer

Yes, Shield is built as an intercepting MCP (Model Context Protocol) proxy, making it seamlessly compatible with MCP clients like Claude Desktop. It integrates transparently into your existing LLM workflow without requiring changes to the client or downstream server.

Question 2

What types of PII can Shield detect and what masking options are available?

Accepted Answer

Shield detects 23 entity types using both regex and NLP engines, including names, emails, SSNs, credit cards, medical IDs, and addresses. It offers flexible masking strategies such as full redaction, partial masking, hashing, pseudo-anonymization, and reversible redaction for controlled data restoration.

Question 3

How does Shield ensure compliance with GDPR and HIPAA?

Accepted Answer

Shield ensures GDPR and HIPAA compliance by proactively redacting PII in real-time, providing comprehensive audit trails, generating automated compliance reports, and offering various masking strategies tailored to privacy regulations. Its proxy architecture prevents sensitive data from ever entering the LLM's context window.

Question 4

Can I customize Shield's detection rules and masking behavior?

Accepted Answer

Absolutely. Shield offers granular control via hot-reloadable TOML configuration files. You can customize per-entity detection rules, set confidence thresholds, apply specific masking strategies for different PII types, and even define allow/deny lists for various tools.

Question 5

What is Shield and what problem does it solve?

Accepted Answer

Shield is a real-time intercepting proxy designed to detect and redact Personally Identifiable Information (PII) from data streams before it reaches large language models (LLMs). It solves the critical problem of sensitive data exposure, protecting organizations from compliance risks and fines.

Shield

Shield

Key Features

Use Cases

Key Features

Use Cases