add-skill
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: Vulnerable to indirect prompt injection. The skill collects untrusted input (skill name, description, trigger phrases) and interpolates it directly into file templates and documentation markers without sanitization or boundary markers.\n
- Ingestion points: Step 1 (skill-name, description, trigger phrases)\n
- Boundary markers: Absent\n
- Capability inventory: File read and write operations (Steps 2, 3, and 4)\n
- Sanitization: Absent\n- [DATA_EXFILTRATION]: Performs file operations using hardcoded absolute system paths (/Users/mikefishbein/Desktop/Vibe Coding/skills/) combined with user-provided variables. This practice reveals information about the host's directory structure and creates a risk of path traversal if inputs are not validated.
Audit Metadata