AI Relationships

A living map of how AI connects — owners, agents, models, MCP servers and the data they reach.

Everything in your AI estate is connected. Relationships make those connections legible so you can read "this AI touches payroll" at a glance instead of counting tools.

The graph#

AuthSpoke composes a graph across your tenant:

Owner → AI Agent → Model
                 → MCP Server → System → Data

Edges are derived from registered agents and MCP servers — an agent's modelProvider, its connectedMcpCount, the systems it reaches, and its exposure flags (sensitiveDataAccess, productionAccess, internetAccess).

Build it from the API#

The graph is rendered client-side from two list endpoints:

curl "https://your-company.authspoke.com/api/v1/ai/agents" -H "Authorization: Bearer $TOKEN"
curl "https://your-company.authspoke.com/api/v1/ai/mcp-servers" -H "Authorization: Bearer $TOKEN"

Because relationships are derived from the same governed records, the map is always current — registering an agent or MCP server updates the graph automatically.

Why it matters#

A relationship-first view answers the questions a security team actually asks: which AI can reach sensitive data, which agents share a model or an MCP server, and what is the blast radius if one is compromised?