Skip to content

Security Posture

The Humanbound security posture is a single 0–100 score (with A–F grade) that summarises an AI agent's severity impact (weighted sum of active findings by severity and status) and coverage effectiveness (ratio of tested threat classes with acceptable pass rates) into one comparable number. It is computed as 100 * (1 - severity_impact) * coverage_effectiveness, viewed via hb posture (current score), hb posture --trends (history), or hb posture --org (aggregated across projects in two dimensions: Agent Security and Quality). The formula is:

posture = 100 * (1 - severity_impact) * coverage_effectiveness
  • Severity Impact: Weighted sum of active findings (by severity and status)
  • Coverage Effectiveness: Ratio of tested threat classes with acceptable pass rates

View Current Posture

hb posture
hb posture --trends

Export as JSON

hb posture --json

View Specific Project

hb posture --project <id>

Organisation-Level Posture (3 Dimensions)

# Aggregate posture across all projects + inventory
hb posture --org

Org posture aggregates two dimensions:

  • Agent Security -- posture across all security-tested projects
  • Quality -- behavioral quality across all tested agents

Include Coverage Breakdown

hb posture --coverage

Posture Grades

Grade Range Description
A 90-100 Excellent security posture with minimal vulnerabilities and comprehensive coverage
B 75-89 Good security posture with minor vulnerabilities or coverage gaps
C 60-74 Fair security posture requiring attention to identified issues
D 40-59 Poor security posture with significant vulnerabilities or coverage gaps
F 0-39 Critical security posture requiring immediate remediation