Developer documentation

Put a security decision in the path of AI actions.

Analyze interactions, test authorized endpoints and operate policies around risky tool use.

Gateway analysis

Send interactions to /api/v1/analyze with X-NovGuard-Key. Requests support text, source context and an optional tool name.

POST /api/v1/analyze
X-NovGuard-Key: ng_...
{"text":"Use the finance tool...","source":"user","tool_name":"finance"}

Policy decisions

NovGuard can produce allow, monitor, require-approval, block or quarantine outcomes. Start in observation mode, inspect events and enforce controls when ready.

Authorized red-team scans

Targets must be verified before scanning. This keeps testing tied to endpoints you control.

Evidence

Keep application, source, tool, decision, severity and remediation with each event so engineers can reproduce and fix the behavior.