shared-security-auth-security

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes an example script ('examples/dependency-security.md') intended for CI/CD pipelines that executes the 'bun audit' command using 'child_process.exec'. This is a routine operation for automated vulnerability scanning within development workflows.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to provide guidance on processing untrusted data safely, creating a potential attack surface if user-provided content is handled without follow-up sanitization. 1. Ingestion points: The skill instructions and examples assume the agent will interact with and review user-provided source code and security configurations as specified in 'SKILL.md'. 2. Boundary markers: The skill utilizes structured markdown tags such as '<critical_requirements>' and '' to isolate core instructions from descriptive content. 3. Capability inventory: Code examples within the skill demonstrate capabilities such as making network requests via 'fetch' ('examples/core.md') and executing system commands for auditing ('examples/dependency-security.md'). 4. Sanitization: The skill explicitly requires and provides implementation details for the sanitization of all user input using 'DOMPurify' before any HTML injection occurs, mitigating XSS and related injection risks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 07:18 PM
Security Audit — agent-trust-hub — shared-security-auth-security