The Agent Skills Directory

[COMMAND_EXECUTION]: The command template in SKILL.md interpolates the and arguments directly into a powershell.exe command string. This pattern is vulnerable to command injection if the underlying platform does not sanitize user inputs, as special characters could be used to escape the quotes and execute arbitrary shell commands.
[COMMAND_EXECUTION]: The provided scripts (remove-template.ps1 and remove-template.py) construct file system paths by joining a base directory with user-supplied object and template names. These scripts do not validate the inputs for directory traversal sequences like '..', which could allow the skill to delete or modify files outside of the intended project scope.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes local XML files which may contain untrusted data, creating a potential surface for indirect prompt injection.
Ingestion points: Both scripts read and parse XML metadata files located in paths derived from the ObjectName argument.
Boundary markers: The XML content is parsed directly without delimiters or instructions to ignore embedded commands.
Capability inventory: The skill has significant capabilities including recursive directory deletion (shutil.rmtree, Remove-Item -Recurse) and file modification.
Sanitization: There is no evidence of validation for the structure or content of the ingested XML files.

template-remove