Understanding Security Flags

SealGate uses three flags to track risk in AI sessions.

Security Flags

📘 Private Data Access

Triggered by: Reading files, querying databases, or accessing internal documents. Risk: Sensitive information could be exposed.

🌐 Untrusted Content Exposure

Triggered by: Fetching web pages or calling external APIs. Risk: The AI could receive malicious instructions (prompt injection).

✉️ External Communication

Triggered by: Sending emails, posting to Slack, or making external API calls. Risk: Confidential data could be exfiltrated.

The Lethal Trifecta

The "Lethal Trifecta" occurs when a session has all three flags active:

Protection: SealGate always tracks these flags, but blocking the action that completes the trifecta is off by default. Until an admin turns on Block lethal-trifecta writes in the Lethal Trifecta Protection card on the Guardrails page, the completing action still goes through and is only recorded. See Settings.

Viewing Flags in the Dashboard

The Sessions view uses colored dots to show active flags:

• 🔵 Blue: Private Data Access
• 🟡 Amber: Untrusted Content Exposure
• 🔴 Red: External Communication

Risk Levels

• Low (green): no External Communication flag, or External Communication on its own
• Medium (amber): External Communication plus either Private Data or Untrusted Content
• High (red): all three flags, the Lethal Trifecta

ACL Levels

Access Control Levels (ACL) provide additional protection:

Enforcement: SealGate automatically blocks high-to-low data flows (e.g., reading SECRET data then posting to a PUBLIC channel). These blocks do not ask for approval - they are prevented by default.

---