Finding categories
The 14 categories cisco-ai-skill-scanner assigns to findings, mapped to the OWASP Agentic Skills Top 10. Each page explains the pattern, shows real examples from the corpus, and lists every affected skill.
| Category | OWASP AST10 | Findings | Skills affected |
|---|---|---|---|
| Data Exfiltration | AST01 Malicious Skills | 745 | 279 |
| Obfuscation | AST08 Poor Scanning | 582 | 38 |
| Policy Violation | AST01 Malicious Skills | 530 | 458 |
| Command Injection | AST01 Malicious Skills | 397 | 152 |
| Unauthorized Tool Use | AST03 Over-Privileged Skills | 212 | 157 |
| Social Engineering | AST01 Malicious Skills | 208 | 167 |
| Transitive Trust Abuse | AST02 Supply Chain Compromise | 151 | 8 |
| Resource Abuse | AST03 Over-Privileged Skills | 91 | 85 |
| Prompt Injection | AST05 Untrusted External Instructions | 65 | 50 |
| Tool Chaining Abuse | AST03 Over-Privileged Skills | 30 | 2 |
| Hardcoded Secrets | AST04 Insecure Metadata | 23 | 4 |
| Supply Chain Attack | AST02 Supply Chain Compromise | 17 | 3 |
| Skill Discovery Abuse | AST04 Insecure Metadata | 6 | 6 |
| Autonomy Abuse | AST09 No Governance | 1 | 0 |