Keydris

A governance layer that checks AI agent authority before actions execute, enforcing centrally assigned policies with allow, approval-required, or reject decisions and auditable decision records. It is for B2B organizations deploying AI agents via MCP and APIs that need to control agent actions at the execution boundary. It is delivered as SaaS with a CLI, background proxy, and harness integration that supplies the authority decision while the customer's boundary enforces it and the tool executes.

Key features

• Authority check before governed actions
• Central policy assignment per agent
• Allow, approval-required, reject decisions
• Short-lived KIT per governed session
• MCP and API integration
• Background proxy for harness wiring
• Auditable decision records with reason
• Human approval workflow for actions