Security Metrics & Monitoring

Security Metrics & Monitoring Framework

This guide provides a comprehensive framework for monitoring, measuring, and improving AI agent skill security through data-driven metrics and analytics.

Key Performance Indicators (KPIs)

Security Metrics

1. Vulnerability Management

Metric: Critical Vulnerabilities Fixed / Total
Target: 100% within 30 days
Tracking: Weekly

CVE Management:
- Published CVEs in skills: [current]
- Fixed this month: [count]
- Pending fixes: [count]
- Average fix time: [days]

2. Malware Detection Rate

Metric: Malicious Skills Detected / Total Skills Published
Target: >95% detection within 30 days of publication
Tracking: Weekly

Samples:
- Total skills scanned: [count]
- Malicious detected: [count]
- False positives: [count]
- Detection accuracy: [percentage]%

3. Security Incident Response

Metric: Mean Time to Respond (MTTR)
Target: <4 hours for HIGH severity
Tracking: Per incident

Response Times:
- CRITICAL: <1 hour (current avg: X min)
- HIGH: <4 hours (current avg: X min)
- MEDIUM: <1 day (current avg: X hours)
- LOW: <1 week (current avg: X days)

4. Publisher Compliance

Metric: Compliant Publishers / Total Active Publishers
Target: >90%
Tracking: Monthly

Compliance Breakdown:
- Signed skills: [percentage]%
- Recent security review: [percentage]%
- Updated documentation: [percentage]%
- No vulnerabilities: [percentage]%

5. User Security Awareness

Metric: Users Who Reviewed Permissions / Total Installers
Target: >70%
Tracking: Quarterly

Survey Data:
- Users reviewing permissions: [percentage]%
- Users verifying publisher: [percentage]%
- Users updating skills: [percentage]%

Dashboard Metrics

Executive Dashboard

For: Leadership, Board, Security Committee

┌─────────────────────────────────────────┐
│ AST10 Security Status - Executive View  │
├─────────────────────────────────────────┤
│                                         │
│ Overall Security Score: 7.8/10         │
│ ████████░░ 78%                          │
│                                         │
│ Key Metrics:                            │
│  • Active Threats: 3 (MEDIUM)          │
│  • Avg Incident Response: 2.2 hrs      │
│  • Publisher Compliance: 87%           │
│  • Skills with Issues: 0.3%            │
│                                         │
│ Trend (30-day): ↑ Improving            │
│                                         │
│ Action Items: 2                         │
│  1. Update 5 outdated policies         │
│  2. Complete Q2 security audit         │
│                                         │
└─────────────────────────────────────────┘

Operations Dashboard

For: Security Operations, Platform Teams

┌──────────────────────────────────────────────┐
│ AST10 Operations - Real-Time Monitoring     │
├──────────────────────────────────────────────┤
│                                              │
│ Skills Published (24h): 127                 │
│ ├─ Scanned: 127 (100%)                     │
│ ├─ Passed: 121 (95%)                       │
│ └─ Flagged: 6 (5%)                         │
│                                              │
│ Active Incidents:                            │
│ ├─ CRITICAL: 0                              │
│ ├─ HIGH: 1 (Incident INC-2026-0042)        │
│ ├─ MEDIUM: 3                                │
│ └─ LOW: 8                                   │
│                                              │
│ Scanning Performance:                        │
│ ├─ Avg Scan Time: 2.3 sec                   │
│ ├─ Malware Detection Rate: 96.2%           │
│ └─ False Positive Rate: 0.8%               │
│                                              │
│ Platform Health:                             │
│ ├─ OpenClaw: ✓ Healthy                     │
│ ├─ Claude Code: ⚠ 2 alerts                 │
│ ├─ Cursor: ✓ Healthy                       │
│ └─ VS Code: ✓ Healthy                      │
│                                              │
└──────────────────────────────────────────────┘

Developer Dashboard

For: Skill Developers, Security Team

┌────────────────────────────────────────┐
│ My Skill Security Status                │
├────────────────────────────────────────┤
│                                        │
│ Published Skills: 12                   │
│ ├─ All Compliant: 10                  │
│ ├─ Minor Issues: 2                    │
│ └─ Critical Issues: 0                 │
│                                        │
│ Latest Scans:                          │
│ ├─ 2026-03-22 10:15: PASS ✓           │
│ ├─ 2026-03-21 14:30: PASS ✓           │
│ ├─ 2026-03-20 09:00: WARN (2 issues)  │
│ └─ 2026-03-19 16:45: PASS ✓           │
│                                        │
│ Security Training Status:              │
│ ├─ Completed Courses: 3/5             │
│ ├─ Certification: AST10-SE Analyst    │
│ └─ Expires: 2027-03-22                │
│                                        │
│ Recommendations:                       │
│ 1. Update deprecated library in skill  │
│ 2. Review and adjust permissions      │
│                                        │
└────────────────────────────────────────┘

Monitoring Infrastructure

Collection Points

┌─ Skill Registry ────────────────────────┐
│ Events:                                  │
│ • Skill published                       │
│ • Skill updated                         │
│ • Skill removed                         │
│ • Download count                        │
│ • User reports                          │
└────────────────────────────────────────┘
       │
       ↓
┌─ Security Scanners ─────────────────────┐
│ AST10-Scanner findings:                 │
│ • Vulnerability detections              │
│ • Malware classifications              │
│ • Permission analysis                  │
│ • Supply chain assessment              │
└────────────────────────────────────────┘
       │
       ↓
┌─ Runtime Monitoring ────────────────────┐
│ Agent execution tracking:               │
│ • Skill invocations                     │
│ • Permission usage                      │
│ • Network activity                      │
│ • File system access                    │
│ • Performance metrics                   │
└────────────────────────────────────────┘
       │
       ↓
┌─ Analytics Platform ────────────────────┐
│ Aggregation & Analysis:                 │
│ • Trend analysis                        │
│ • Anomaly detection                     │
│ • Risk scoring                          │
│ • Correlation analysis                  │
└────────────────────────────────────────┘
       │
       ↓
┌─ Visualization & Alerting ──────────────┐
│ Dashboards & Notifications:             │
│ • Real-time dashboards                  │
│ • Automated alerts                      │
│ • Reports & trends                      │
│ • Escalation workflows                  │
└────────────────────────────────────────┘

Data Collection

import metrics

class SkillMetricsCollector:
    def __init__(self):
        self.metrics = metrics.MetricsClient()
    
    def record_skill_event(self, event_type, skill_id, metadata):
        """Record skill lifecycle events"""
        self.metrics.increment(
            'skill.events',
            tags={'type': event_type, 'skill_id': skill_id}
        )
        self.metrics.gauge(
            'skill.downloads',
            metadata.get('download_count'),
            tags={'skill_id': skill_id}
        )
    
    def record_scan_result(self, scan_result):
        """Record security scan results"""
        self.metrics.increment(
            'scans.total',
            tags={'status': scan_result.status}
        )
        self.metrics.histogram(
            'scans.duration_ms',
            scan_result.duration_ms
        )
        
        for finding in scan_result.findings:
            self.metrics.increment(
                'findings',
                tags={'severity': finding.severity}
            )
    
    def record_incident(self, incident):
        """Record security incidents"""
        self.metrics.increment(
            'incidents',
            tags={'severity': incident.severity}
        )
        self.metrics.histogram(
            'incident.response_time_minutes',
            incident.response_time
        )

Alerting Rules

Critical Alerts

Alert: MaliciousSkillDetected
Condition: Malware confidence > 90%
Action: 
  - CRITICAL incident created
  - Skill immediately flagged
  - Security team paged
  - Incident response activated

High Severity Alerts

Alert: UnexpectedPermissions
Condition: Skill requests unusual permission combination
Action:
  - HIGH incident created
  - Manual review queued
  - Publisher notified
  - Enhanced monitoring enabled

Medium Severity Alerts

Alert: VulnerabilityFound
Condition: CVE published affecting skill dependency
Action:
  - MEDIUM incident created
  - Affected publishers notified
  - Patch available communication sent
  - Tracking enabled

Low Severity Alerts

Alert: ComplianceGap
Condition: Publisher documentation outdated (>90 days)
Action:
  - LOW priority ticket created
  - Publisher sent reminder email
  - Tracking enabled

Trend Analysis

Quarterly Report Example

Q1 2026 Security Report
========================

Executive Summary:
- 1,247 new skills published
- 12 malicious skills detected (0.96%)
- 23 security incidents resolved
- 87% publisher compliance

Trend Analysis:
- Malware Detection: ↑ 15% (improved detection)
- Response Time: ↓ 25% (faster response)
- Publisher Compliance: ↑ 5% (better education)
- User Awareness: ↑ 12% (more training)

Key Findings:
1. Supply chain attacks increasing (trend analysis)
2. Permission escalation most common vulnerability
3. Platform A has 3x more issues than B
4. Certain publisher type has 5x incident rate

Recommendations:
1. Increase supply chain scanning
2. Add permission validation warnings
3. Platform A needs hardening
4. New publisher education program

Anomaly Detection

Machine Learning Models

Model 1: Skill Behavior Analysis
- Detects unusual permission patterns
- Identifies obfuscation techniques
- Flags suspicious code structures
- Accuracy: 94.2%

Model 2: Publisher Assessment  
- Analyzes publisher history
- Detects risky publisher patterns
- Predicts future incidents
- Accuracy: 87.6%

Model 3: Network Traffic Analysis
- Detects unusual outbound connections
- Identifies data exfiltration
- Flags C2 communication
- Accuracy: 96.1%

Alert Confidence Scoring

Score: 0-100 (higher = more confidence)

85-100: CRITICAL - Immediate action
70-84:  HIGH - Urgent investigation  
50-69:  MEDIUM - Schedule review
25-49:  LOW - Monitor
0-24:   INFO - Log only

Examples:
- Skill connects to known C2: 98/100 → CRITICAL
- Suspicious permission combo: 62/100 → MEDIUM
- Rare library usage: 35/100 → LOW

Reporting

Automated Reports

Daily Security Summary

Date: 2026-03-22
Skills Published: 127
Security Scans: 127 (100%)
Passed: 121 (95.3%)
Issues Found: 6 (4.7%)
Malware Detected: 0 (0%)
Incidents Opened: 2
Incidents Closed: 1
New Malicious Skills: 3 (↓ 40% from previous week)
Publisher Violations: 5 (↑ 25%)
User Complaints: 12 (↓ 15%)
Security Training: 34 enrolled (↑ 5%)

Monthly Report

  • 30-day trends
  • Emerging threats
  • Successful mitigations
  • Performance metrics
  • Recommendations

Continuous Improvement

Metrics Review Cycle

Weekly:
  - Alert review  
  - Incident metrics
  - Detection rates

Monthly:
  - Dashboard review
  - Trend analysis
  - Alert tuning

Quarterly:
  - Full metric assessment
  - Model retraining
  - Strategy adjustment

Annually:
  - Goals review
  - Program evaluation
  - Long-term trends

Benchmarking

Industry Benchmarks (2026):
- Malware detection rate: 85-95%
- Incident response time: 2-6 hours
- Publisher compliance: 70-85%
- False positive rate: 1-3%

Our Performance:
- Detection rate: 96.2% ✓
- Response time: 2.1 hours ✓
- Compliance: 87% ✓
- False positives: 0.8% ✓

Status: Above industry average

Tools & Platforms

Collection:
- Prometheus (metrics)
- Elasticsearch (logs)
- Jaeger (tracing)

Analysis:
- Splunk / ELK
- Grafana
- Datadog

Alerting:
- PagerDuty
- Opsgenie
- Slack

Visualization:
- Grafana
- Kibana
- Splunk

Metrics framework updated: March 2026. Review quarterly with leadership.


Example

Put whatever you like here: news, screenshots, features, supporters, or remove this file and don’t use tabs at all.


Leadership & Founding Members

Project Leadership

Current Leaders

Ken Huang

Ken Huang

Hammad Atta

Hammad Atta

Fabio Cerullo

Fabio Cerullo

Aonan Guan

Aonan Guan

Bhavya Gupta

Bhavya Gupta

Niv Hoffman

Niv Hoffman

Iftach Orr

Iftach Orr

Akram Sheriff

Akram Sheriff

AIVSS Distinguished Review Board

The OWASP AIVSS project’s Distinguished Review Board comprises world-renowned cybersecurity leaders, former government officials, and industry pioneers who provide strategic guidance and expert oversight for the AI Vulnerability Scoring System framework. We thank them for their guidance, several of whom have also supported this project’s work.

Rob Joyce

Rob Joyce

Advisor to PwC and OpenAI, Former Special Assistant to the President and Cybersecurity Coordinator

Jason Clinton

Jason Clinton

Deputy CISO, Anthropic

Amy R. Steagall

Amy R. Steagall

Chief Information Security Officer, Stanford University

Martin Stanley

Martin Stanley

AI Risk Management Framework Lead, NIST

Apostol Vassilev

Apostol Vassilev

Research Supervisor, NIST

Andrew Coyne

Andrew Coyne

CISO, Banner Health, Former CISO, Mayo Clinic

Kevin Rocque

Kevin Rocque

Managing Director/Executive Vice President, Global Technology Risk Officer, TD Bank

Jeff Williams

Jeff Williams

Former Global OWASP Chair, Founder and CTO, Contrast Security

Michael Tran Duff

Michael Tran Duff

University Chief Information Security and Data Privacy Officer, Harvard University

Emil Bender Lassen

Emil Bender Lassen

Standards Lead, AIUC-1

Agentic Skills Top 10 Founding Members

Founding members of the OWASP Agentic Skills Top 10 project itself — project leads, co-leads, and additional contributors — listed alphabetically. Several also contribute to the sibling OWASP AIVSS project listed above.

Ken Huang

Ken Huang

Project Lead, Agentic Skills Top 10

Hammad Atta

Hammad Atta

Co-Lead, Agentic Skills Top 10

Manish Bhatt

Manish Bhatt

Security Researcher, AWS

Fabio Cerullo

Fabio Cerullo

Co-Lead, Agentic Skills Top 10

David Girard

David Girard

Senior Director, AI Security & AI Alliances, Trend Micro

Aonan Guan

Aonan Guan

Co-Lead, Agentic Skills Top 10

Bhavya Gupta

Bhavya Gupta

Co-Lead, Agentic Skills Top 10

Pamela Gupta

Pamela Gupta

Founder & CEO, OutSecure / Trusted AI

Idan Habler

Idan Habler

Staff AI/ML Security Researcher, Intuit

Niv Hoffman

Niv Hoffman

CTO, Air Security

Charles Iheagwara

Charles Iheagwara

AI/ML Security Leader, AstraZeneca

Sushmitha Janapareddy

Sushmitha Janapareddy

Director - Security Integrations, American Express

Edward Lee

Edward Lee

Vice President, Lead AI Security, JP Morgan

KJ Lian

KJ Lian

Senior Manager, Data & AI (Public Sector), AWS

Vineeth Sai Narajala

Vineeth Sai Narajala

Application Security, AWS

Iftach Orr

Iftach Orr

Co-Lead, Agentic Skills Top 10

Kanna Sekar

Kanna Sekar

Cyber Security, Google

Akram Sheriff

Akram Sheriff

Co-Lead, Agentic Skills Top 10

Dennis Xu

Dennis Xu

Research VP, AI, Gartner

OWASP AIVSS Founding Members

The OWASP AIVSS (Agentic AI Vulnerability Scoring System) project is a sibling OWASP initiative focused on scoring the severity of agentic AI vulnerabilities. Its founding members are recognized here as OWASP founding members in the agentic AI security space; many of them have also contributed directly to the Agentic Skills Top 10 project’s research and review process.

Sunil Agrawal

Sunil Agrawal

Chief Information Security Officer, Glean

David Ames

David Ames

Partner, PwC

Michael Bargury

Michael Bargury

Founder and CTO, Zenity

Joshua Beck

Joshua Beck

Application Security Architect, SAS

Manish Bhatt

Manish Bhatt

Security Researcher, Amazon Kuiper Security

Mark Breitenbach

Mark Breitenbach

Security Engineer, Dropbox

Anat Bremler-Barr

Anat Bremler-Barr

Professor of Computer Science, Tel Aviv University

Siah Burke

Siah Burke

HIPAA Security Officer, Siah.ai

David Campbell

David Campbell

AI Security, Scale AI

Ying-Jung Chen

Ying-Jung Chen

AI safety researcher, PhD, Georgia Institute of Technology

Anton Chuvakin

Anton Chuvakin

Security Solution Strategy, Google

Jason Clinton

Jason Clinton

CISO, Anthorphic

Adam Dawson

Adam Dawson

Staff AI Security Researcher, Dreadnode

Leon Derczynski

Leon Derczynski

Principal Research Scientist, NVIDIA

Walker Lee Dimon

Walker Lee Dimon

AI Security Researcher, MITRE

Marissa Dotter

Marissa Dotter

AI Security Researcher, MITRE

Dan Goldberg

Dan Goldberg

ISO Market Lead, Omnicom

David Haber

David Haber

CEO, Lakera

Idan Habler

Idan Habler

Staff AI/ML Security Researcher, Intuit

Jason Haddix

Jason Haddix

Founder, Arcanum Information Security

Keith Hoodlet

Keith Hoodlet

Director of AI/ML & AppSec, Trail of Bits

Ken Huang

Ken Huang

AIVSS Project Lead, OWASP

Chris Hughes

Chris Hughes

CEO, Aquia

Charles Iheagwara

Charles Iheagwara

AI/ML Security Leader, AstraZeneca

Krystal Jackson

Krystal Jackson

Researcher, Center for Long-Term Cybersecurity, UC Berkeley

Sushmitha Janapareddy

Sushmitha Janapareddy

Director - Security Integrations, American Express

Rob Joyce

Rob Joyce

Former Cybersecurity Director of NSA, Advisor to PwC, PwC

Diana Kelley

Diana Kelley

CISO, Noma Security

Prashant Kulkarni

Prashant Kulkarni

Lead AI Security Research Engineer, Google Cloud

Mahesh Lambe

Mahesh Lambe

Founder, MIT, Unify Dynamics

Edward Lee

Edward Lee

Vice President, Lead AI Security, JP Morgan

Nate Lee

Nate Lee

CEO, Cloudsec.ai

Vishwas Manral

Vishwas Manral

CEO, Precize.ai

Daniela Muhaj

Daniela Muhaj

Executive-in-Residence for Research & Development, AI 2030

Vineeth Sai Narajala

Vineeth Sai Narajala

Application Security, AWS

Om Narayan

Om Narayan

AI Security Researcher, AWS

Varun Pant

Varun Pant

Engineering and Product Leader, AI applications at the Automated Reasoning Group, AWS

Advait Patel

Advait Patel

Senior Site Reliability Engineer (DevSecOps + Cloud + AIOps), Broadcom, IEEE

Alex Polyakov

Alex Polyakov

CEO, adversa.ai

Ramesh Raskar

Ramesh Raskar

Professor & Director, MIT Media Lab

Ron F. Del Rosario

Ron F. Del Rosario

VP-Head of AI Security, SAP

Tal Shapira

Tal Shapira

Co-Founder & CTO, Reco AI

Akram Sheriff

Akram Sheriff

Senior AI/ML Software Engineering Leader, Cisco

Samantha Siau

Samantha Siau

Security and Compliance, Anthropic

Kevin Simmonds

Kevin Simmonds

Partner on AI Offensive Security, PWC

Martin Stanley

Martin Stanley

NIST AI RMF Lead, Independent

Omar A. Turner

Omar A. Turner

General Manager of Security, Microsoft

Apostol Vassilev

Apostol Vassilev

AI Research Team Supervisor, NIST

Matthew Versaggi

Matthew Versaggi

AI Fellow, White House Presidential Innovation Fellow

David Webb

David Webb

Agency Cybersecurity Officer, Cybersecurity and Infrastructure Security Agency

Dennis Xu

Dennis Xu

Research VP, AI, Gartner

Xiaochen Zhang

Xiaochen Zhang

Executive Director and Chief Responsible AI Officer, AI 2030

Recognition

We extend our gratitude to all founding members who have contributed to establishing this crucial framework for AI security assessment. Their vision and dedication have been instrumental in shaping the Agentic Skills Top 10 project.

Get Involved

Interested in contributing to the Agentic Skills Top 10 project? We welcome new contributors and leaders. Please see our Contribution Guidelines for more information on how to get involved.