Policy at the MCP boundary
Evaluate discovery, request, and response phases before a tool, prompt, resource, or generated capability reaches a model context.
Enterprise Governance
Enterprise-grade control for AI agents, tools, prompts, and shared context.
ThisMCP™ puts repeatable controls at the protocol boundary. The model may request an operation, but the gateway decides whether it is visible, executable, redacted, approval-gated, or blocked.
Give security and operations teams controls that do not rely on prompt instructions alone.
- Policy can hide, block, redact, downgrade, route, quarantine, defer, or require approval.
- Generated capabilities can be reviewed before promotion into a group bundle or task capsule.
- Audit records can correlate discovery output, policy versions, approvals, credential selection, and runtime outcomes.
Gate 1outside model
Discovery
Show only allowed tools, prompts, and resources.
Gate 2outside model
Request
Check arguments, identity, budget, and approval state.
Gate 3outside model
Response
Redact, summarize, log, or block sensitive results.
Allow
Approve
Block
Benefits
What changes for the team.
Credentials stay shielded
Keep upstream credentials out of the agent while still selecting the eligible group, member, service, or task credential at call time.
Evidence for every decision
Use approvals, budgets, temporal windows, manifests, and audit trails to make higher-consequence AI actions governable.
Next Step