Sandbox — Explore with sample data from Acme Corp
Start Free →
Agents/hr-screening-agent

HR Screening Agent

Pre-screens job applications and ranks candidates. Processes ~300 applications/week.

5d ago
activehigh risk

Insights

CRITICAL

Remediation critically behind schedule

HR Screening Agent is classified as high risk with only 17% remediation complete. At current pace, compliance will not be achieved before the EU AI Act deadline in 74 days.

View action plan
CRITICAL

1 critical incident unresolved

HR Screening Agent has 1 P1 incident open. The oldest has been unresolved for 14 days. Critical incidents indicate customer-facing impact or safety issues.

View incidents
CRITICAL

Production agent not reporting

HR Screening Agent is a production agent that has not sent a heartbeat in 6 days. This could indicate a silent failure affecting end users.

Check agent status
WARNING

Error rate at 27.8%

HR Screening Agent has a 27.8% error rate over the last 30 days (59 errors out of 212 requests). Error rates above 5% indicate potential reliability issues that could trigger compliance concerns.

View metrics
WARNING

Cache efficiency at 17.7% — below 60% threshold

HR Screening Agent has a 17.7% cache hit rate over the last 30 days (49,089 hits, 228,778 misses). Low cache efficiency means repeated prompt processing costs — optimizing prompt prefixes can reduce costs by 50–90%.

WARNING

No token budget configured

HR Screening Agent is a production agent with no spend thresholds. Without a runaway kill-switch, a prompt injection or logic bug could generate unbounded costs. Create a Token Strategy Charter to set guardrails.

Create charter

Details

Agent ID
hr-screening-agent
Provider
openai
Model
gpt-4o
Owner
maria.garcia@acme.com
Department
Human Resources
Environment
production
Registered
5/14/2026

Compliance Trends

3 assessments across 2 frameworks(2/7/2026 – 5/5/2026)

Overall Stable
EU AI Act New
High Risk2 assessmentsLast: 5/5/2026
No score
GDPR AI Processing Assessment New
1 assessmentLast: 5/3/2026
42%

Cascading Risk

high
1
Upstream
0
Downstream
0
Blast Radius

Bias Monitoring

HIGH ALERT
56/100
poor fairness
Trend (last 10)
70

Based on 983 decisions · 5/14/2026

Demographic Parity43/100
Equalized Odds58/100
Calibration67/100

Action required

Bias score below acceptable threshold. Review decision outcomes across demographic groups, audit training data for representation gaps, and consider re-calibrating model outputs.

Metrics (Last 30 Days)

Tags

high-riskemploymentdecisions-affecting-people

Incidents

View all
p1Bias detected in candidate screening — female applicants scored 15% lower
open5/7/2026