review-skill-sflabs

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to inspect local file system directories, such as ~/.local/share/cortex/, to identify currently installed bundled skills for catalog comparison.
  • [EXTERNAL_DOWNLOADS]: Fetches skill catalog metadata from the vendor's official documentation at docs.snowflake.com to ensure the readiness check is performed against the most current information.
  • [PROMPT_INJECTION]: Because the skill reads and processes the content of untrusted local files (the skill being audited), it is exposed to indirect prompt injection. Adversarial instructions within the audited files could attempt to influence the agent's logic, bypass policy scans, or manipulate the generated report.
  • [DATA_EXFILTRATION]: Performs web_search operations using metadata fragments (name and description) from the audited skill. This is a primary feature for detecting public duplicates but involves sending fragments of the local skill content to a search provider.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 06:35 PM
Security Audit — agent-trust-hub — review-skill-sflabs