image-upload
Warn
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to use shell utilities (
lsandcurl) with file paths provided directly by the user. If the agent does not properly sanitize the path input, this creates a risk of command injection where an attacker could use shell metacharacters (e.g., semicolons, pipes, or backticks) to execute arbitrary commands. - [EXTERNAL_DOWNLOADS]: The skill is configured to automatically fetch a temporary API key from
https://images.labnocturne.com/keywhen a local key is not provided in the environment. This constitutes an external network request to a vendor-controlled endpoint to retrieve runtime configuration. - [DATA_EXFILTRATION]: The primary function of the skill is to transmit local data to a remote server (
images.labnocturne.com). The skill includes a mitigation by enforcing an extension whitelist (jpg, jpeg, png, gif, webp, svg) to prevent the upload of sensitive text-based configuration or credential files, but the core capability of transferring local data to an external destination remains a notable security behavior.
Audit Metadata