expertise-file-design

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • PROMPT_INJECTION (SAFE): No instructions attempting to override agent constraints or bypass safety filters were detected.
  • DATA_EXFILTRATION (SAFE): No network operations, hardcoded credentials, or sensitive file path access patterns were found.
  • REMOTE_CODE_EXECUTION (SAFE): The skill does not download or execute remote scripts and contains no package management files.
  • NO_CODE (SAFE): The skill consists exclusively of markdown documentation and YAML templates without any executable scripts or binary files.
  • INDIRECT_PROMPT_INJECTION (SAFE): Although the skill provides templates for data the agent will read, it does not involve untrusted data interpolation or missing boundary markers that would facilitate injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:17 PM