MCP SERVER

Read-only MCP over the tenant's own risk telemetry

An approved external AI client can query scoped tenant data with a token your team issues. Calls are audited, and tokens can be revoked.

Structured context for approved AI workflows

MCP makes approved reads machine-consumable. Action and authorization stay in your own systems.

Your MCP client

Bearer token

  • Scope A
  • Scope B
  • Scope C: not in this token

The narrowest useful scope for each integration.

The tenant the token was issued to

Read-only tools

  • Policies
  • Rules
  • Alerts
  • Risk distribution

Every tool is a read. Calls are recorded, including denied calls.

Another tenant

Out of reach: the token is tenant-bound.

  • Read-only: no configuration changes, no action on traffic
  • Responses are data to interpret, not instructions
  • Can be out of date after configuration or traffic changes
Scoped and tenant-bound: a token reaches read-only tools for its own tenant, and nothing beyondIllustration with example labels, not customer data.
  • Read-only tools

  • Scoped, tenant-bound tokens

  • Audited calls

  • Controlled token lifecycle

From approval to a reviewed answer

  1. Console

    Issue a scoped token

    An authorized operator creates a token with the narrowest useful read scopes.

  2. Your MCP client

    Present the token

    Bearer token

    An MCP-compatible client, set up with the details from your tenant provisioning.

  3. Noxtica

    Read-only tools

    A supported read

    Reads the tenant's own risk telemetry within the token's scopes. Calls are audited, including denied calls.

  4. Your team

    Human review or your automation

    Data, not instructions

    The client interprets the response. Decisions and actions stay in your own systems.

How a scoped token turns into a read-only answer your team reviews
  1. Confirm access for the tenant

  2. Issue a scoped token

  3. Connect a client and review

Boundaries and responsibilities

  • Controlled availability. MCP access follows the tenant's plan tier and operator gates.
  • The tools for tenant analytics are read-only. They do not change Noxtica configuration or act on traffic.
  • Responses are data for the client to interpret, not instructions.
  • Scores and risk levels in responses are probabilistic and can be wrong. Your team decides what to do, and is responsible for the client and model it connects.