exec-summary

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill explicitly declares the use of powerful system tools including 'bash', 'git', and 'File ops' to interact with project files and environment.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection attacks because it processes untrusted data with high-privilege capabilities.
  • Ingestion points: Business content and data reviewed during 'Step 1: Intake and Analysis' in SKILL.md.
  • Boundary markers: There are no defined delimiters or instructions to ignore embedded commands within the business inputs.
  • Capability inventory: The skill maintains access to the bash shell, git version control, and project file system operations.
  • Sanitization: The skill lacks any input validation or sanitization mechanisms to filter malicious instructions from the analyzed content.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 02:51 PM
Security Audit — agent-trust-hub — exec-summary