Products | DIMAGGI AI

Tool Guard

Tool Guard is DIMAGGI AI's runtime policy enforcement engine for AI agents. It intercepts every tool call, evaluates it against your policy library, and makes a real-time ALLOW / DENY / ESCALATE decision.

Shadow Mode

Deploy Tool Guard in shadow mode to observe all agent tool calls without blocking — understand your risk surface before enforcing policies.

Enforcement Mode

Activate enforcement to block policy violations in real time. Every denied call produces a signed, hash-chained audit record.

Policy types

• Scope-lock: restrict agents to approved tool lists
• PII boundary: redact sensitive fields before external transmission
• Confidence gate: escalate low-confidence decisions to humans
• Contraindication / semantic safety: LLM-powered deep evaluation