AgentProof – Stability Scan Sample

Date · May 2025

Executive Summary

Overall Score

86 / 100

Agents Covered

4 agents · 32 prompts

Critical Findings

2

Cost Forecast 30d

$1 215

Security & Behaviour Radar

Radar chart

Key Findings

CheckStatusExample
🔐 API-Key Leak2 highHard-coded sk-live-••• in util.py
⚡ Rate-Limit Heuristic1 medLooping LLM call w/ no delay
🛡️ Input Sanitization3 medUser input f-string into prompt
🧬 Fingerprint Drift1 medAgent-B response deviates >25 %
🔊 Echo Leak1 highUser email reflected in reply
🚨 Unauthorized Call Pattern1 medOpenAI call from non-whitelisted IP
📈 Token Spike Anomaly1 med+480 % tokens vs baseline
Latency SLO✓ within 900 ms—
Hover redacted cells to preview full detail.

High-Impact Fixes (excerpt)

  1. Move secrets to env vars; rotate exposed keys.
  2. Throttle vector-search loop with exponential back-off.
  3. Escape user-supplied content before prompt insertion.
  4. Lock agent replies to whitelisted domains to prevent echo leak.

Projected ROI After Fixes

Spend After Fix

$795 / 30 d

Est. Savings

$420 / mo

Payback

< 1 week

Next Steps

Go continuous—daily scans, live drift alerts, certification badge.

$299 / mo

Monthly monitoring
& badge refresh

Start monthly

$2 499 / yr

2 months free

Start annual

Need SOC-ready certification?

View the $999 Monitoring + Certification tier →

Generated by AgentProof Engine