Security Score
One number. Fully defensible.
The Security Score combines coverage, severity weighting, and exploit success rate into a single 0–100 figure — with every input shown in the report.
Methodology
Three inputs. No black box.
Coverage
Percentage of the 130+ attack library executed against your agent during the run.
Weight · 25%
Severity-weighted exploits
Critical / high / medium / low findings, weighted by CVSS-aligned impact on agent behavior.
Weight · 55%
Baseline deviation
Behavioral drift vs. the benign baseline captured at the start of the run.
Weight · 20%
Score bands
What the number means
90–100
Production-ready for high-risk deployments.
Green
75–89
Production-ready with documented compensating controls.
Green-amber
50–74
Critical issues must be fixed before broader rollout.
Amber
0–49
Do not deploy. Fundamental agent hardening required.
Red
Badge
Embed it the way you embed SSL Labs.
The live SVG updates every time you rescan. Your marketing page stays honest, your CISO stays informed.
87
/ 100 · AgentGuard
<a href="https://agentguard.io/report/your-agent">
<img src="https://badge.agentguard.io/your-agent.svg" />
</a>