security-audit-pro

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
  • NO_CODE (SAFE): The skill documentation in SKILL.md references two specialized utility scripts (scripts/simulate-leak.ts and scripts/extract-audit-report.py) that are not provided in the skill package. This omission prevents a technical security review of the executable components but does not pose a direct threat in the current context.
  • PROMPT_INJECTION (LOW): The skill presents an Indirect Prompt Injection surface (Category 8) because it is designed to analyze external data sources such as database schemas and policies. 1. Ingestion points: Analysis of database policies and Convex function permissions (SKILL.md, Protocol Step 2). 2. Boundary markers: Absent; there are no explicit instructions for the agent to disregard instructions embedded within the audited code or schema comments. 3. Capability inventory: The skill references script execution capabilities. 4. Sanitization: Not implemented in the provided prompt logic.
  • DATA_EXPOSURE (SAFE): No hardcoded credentials, sensitive file paths, or private keys were found. All SQL and TypeScript examples use standard security patterns and generic placeholders.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:29 PM