XSOC AIDA-Guard Range

Agent action and identity governance, built on the Nexus Identity Engine

Bring your own agent

Test your agent against a Guard that XSOC hosts.

AIDA-Guard decides every tool call an AI agent makes against a capability token: which tools it may call, with which arguments, how many times and for how long. A persuaded agent can try anything; the token decides what it can do. Every decision is recorded on a Providence chain that you can verify yourself, offline.

The demo

A guided session with XSOC

XSOC presents five scenes on the hosted console over a video call, by invitation. Ask your XSOC contact to arrange one.

The Range

Hands-on, with your own client

Connect the agent you already use, run the test card yourself against a sandboxed tool set, and take your evidence with you. Access is by approved request.

Request access Check a request

What a Range session looks like

  1. Request access. XSOC reviews each request, and an approved request gives you an invitation.
  2. Enroll this browser as your device on the Nexus Identity Engine, with a passkey. The device key is derived from your passkey each time you use it; your agent never sees it.
  3. Connect your MCP client to your own Guard endpoint: Claude Desktop, Cursor, OpenClaw, or any client that can send a request header.
  4. Work through the test card: granted calls go through, and every attempt outside the grant is refused and recorded.
  5. Download your chain with its anchors and checkpoints, and verify it on your own machine.