digikam-media-pack-copy
Warn
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
find <dest> -type f -deletewith a user-provided destination path. This presents a potential command injection or path traversal risk if the agent does not strictly validate or quote the<dest>variable before execution. - [COMMAND_EXECUTION]: User-supplied tag names are directly interpolated into SQL strings (
WHERE name IN (...)). This creates a surface for SQL injection where a crafted tag name could be used to manipulate database queries beyond the intended scope. - [REMOTE_CODE_EXECUTION]: The workflow involves generating multiple Python scripts at runtime and executing them using
uv run. This dynamic execution pattern is used for core logic (shuffling and timestamp randomization). While the logic is local, executing code generated based on potentially untrusted metadata (file paths or tags) introduces a risk of subversion if that data contains malicious payloads. - [COMMAND_EXECUTION]: The skill performs direct read access to a sensitive personal file (
~/Pictures/digikam4.db). Although this is the primary purpose of the skill, it exposes structured personal metadata (tags, locations, file structures) to the agent's processing context.
Audit Metadata