The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local validation script (cf-validate) to verify configuration integrity after edits are performed. This execution uses subprocess.run (Python) and powershell.exe (PowerShell) to call a script that is part of the same trusted toolset.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied file paths and JSON definition files containing modification instructions. It mitigates potential injection risks by using standard XML libraries (lxml.etree and XmlDocument) and implementing explicit escaping for user-supplied string values inserted into the XML structure.
[DYNAMIC_EXECUTION]: The Python implementation uses __import__ to dynamically load the standard re module and re.compile for regular expression initialization. These are standard programming practices used for performance and modularity, and do not involve executing untrusted code or external payloads.
[SAFE]: The skill incorporates a 'support-guard' mechanism that verifies the support status of 1C objects by inspecting ParentConfigurations.bin and project configuration files (.v8-project.json) before allowing modifications. This serves as a significant safety feature to protect developers from accidentally breaking managed vendor updates.

cf-edit