raisindb-file-uploads

Pass

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes JavaScript boilerplate for server-side functions (triggers and functions) designed to run within the RaisinDB platform's execution environment. These functions are intended for legitimate asset processing such as generating thumbnails and extracting metadata.
  • [DATA_EXFILTRATION]: Access to binary files is restricted to signed URLs generated via a server-side HMAC process, which is a security best practice to ensure files are not publicly exposed and access is time-limited.
  • [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by suggesting the use of AI models to analyze user-uploaded content for metadata generation. However, it mitigates potential secondary attacks (like SQL injection) by using parameterized queries when saving results back to the database.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 3, 2026, 02:10 AM