The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and processes filenames/directory paths from local storage, which are external ingestion points.
Ingestion points: storage.list output containing file and directory names.
Boundary markers: The instructions lack delimiters or warnings to ignore instructions that might be embedded in filenames.
Capability inventory: The skill has access to execute_command and state-changing API operations like transfer.file.
Sanitization: There is no requirement for the agent to sanitize or validate the content of filenames before classification or action.
[REMOTE_CODE_EXECUTION]: The workflow specifies the use of scripts/mp-downloader.py via the execute_command tool. While this is described as a 'fixed' helper, the code for this script is not present in the skill package, making it an unverified external dependency that performs shell execution.

organize-files