Risk & Trust

Every AI identity carries two continuously computed scores — Risk (0–100) and Trust (0–100). They are computed server-side and returned ready to render; you never submit them.

Risk — blast radius#

Risk reflects what an agent can reach and do. It rises with:

• Production access and sensitive-data access (the largest contributors)
• Internet access and external tool usage
• The ability to spawn other agents
• The number of connected systems and connected MCP servers
• Business criticality

Scores are clamped to 0–100. A high-blast-radius agent reads high on risk regardless of how well it's governed.

Trust — governance posture#

Trust reflects how well an agent is governed. It rises with strong authentication (MTLS > OAUTH > API_KEY > NONE) and a healthy lifecycle (APPROVED adds trust; DISCOVERED, DEPRECATED, RETIRED subtract it), and falls for discovered/shadow agents and high uncontrolled risk.

Bands#

The overview groups agents into bands so you can triage:

Read scores#

# per agent
curl "https://your-company.authspoke.com/api/v1/ai/agents/$ID" -H "Authorization: Bearer $TOKEN"
# tenant rollup, bands, averages and top-risky agents
curl "https://your-company.authspoke.com/api/v1/ai/overview" -H "Authorization: Bearer $TOKEN"

Because the score function is transparent, it's also explainable — you can always point to the factors that drove a given agent's risk.