AgentGuard Jev Prompt Injection Guardrail
Put a prompt-injection guardrail in front of any Boomi Agentstudio agent. A web service API screens every user message with TypeSafe Jev before it reaches the agent, routes each turn to BLOCK, REVIEW or PASS against configurable thresholds, calls the agent only when the message is not blocked, and logs every verdict. Fails closed if Jev is unavailable. Includes a sample HR Policy Helper agent.
Solution Details
- Difficulty Level
- INTERMEDIATE
- Solution Type
- AI Agent Recipe
- Author
- Boomi
- Published on
- Nov 19, 2026
- Last updated on
- Nov 19, 2026
Get Started
Get StartedKey Features
- Screens every user message with TypeSafe Jev for prompt injection before it reaches the agent
- Routes each turn to BLOCK, REVIEW or PASS using configurable thresholds
- Fails closed: any error or missing Jev score blocks the turn
- Audit log entry for every verdict, answered or blocked
- JSON web service API in front of the agent, with a sample HR Policy Helper agent included
Applications Required
- REST Client
- Web Services Server
- Jev