1 Overview
Agent Rankings lists every assessed agent across several views — each with a per-component heatmap breakdown. Every section has its own independent filters. Scoring follows the AIRQ methodology.
Every agent has a page with a full analytical profile — click any agent name in the rankings below.
Compare Agents lets you pick up to three agents and view headline scores, quadrant placement, and full per-component breakdowns side by side.
AIRQ Score Ranking measures composite AI Risk Appetite for each agent — Attack Surface, Blast Radius, and Defense Controls combined into one score that rewards defended capability and penalizes both unprotected exposure and underdeveloped capability.
Attack Surface Ranking measures exposure across ten lifecycle surfaces: model input, tool use, code execution, file system, network, credentials, data stores, third-party integrations, deployment access, and autonomous action.
Blast Radius Ranking measures damage potential across six components: code execution, file system, network reach, credentials, autonomous action, and deployment access.
Defense Controls Ranking measures vendor control coverage across five areas: input guardrails, execution isolation, action controls, output guardrails, and monitoring & audit.
2 Compare Agents
Select up to three agents to compare headline scores, quadrant placement, and full Attack Surface, Blast Radius, and Defense Controls breakdowns. Compare within the same agent class when possible — cross-class comparisons can be category errors.
3 AIRQ Score Ranking
Composite score aligned with AI Risk Appetite — how much defended capability the product supports. Defense investment lifts the score; attack surface exposure lowers it.
4 Attack Surface Ranking
Attack Surface is a weighted sum of ten lifecycle surfaces. The table shows each agent's per-surface score so you can see where exposure concentrates, not just how large it is. Rows sort by Attack Surface descending — most exposed first.
5 Blast Radius Ranking
Blast Radius is a weighted sum across six damage components — code execution, file system, network reach, credentials, autonomous action, and deployment access. The table shows each agent's per-component score so you can see what kind of compromise lands the hardest. Rows sort by Blast Radius descending — most damaging first.
6 Defense Controls Ranking
Defense Controls is a sum across five vendor controls — input guardrails, execution isolation, action controls, output guardrails, and monitoring & audit. The table shows each agent's per-control score so you can see which controls are present and which are missing. Rows sort by Defense Controls ascending — least defended first.