The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for installing local Python wheel files using pip and dynamically loading the Siemens scripting module by modifying the system path (sys.path) based on the TIA_SCRIPTING environment variable. These patterns are standard for manual SDK integration but involve executing code from variable filesystem paths.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external automation data (XML, SimaticML, CAx, and CFC charts) into TIA Portal projects. This creates a surface where malicious instructions could be embedded within engineering artifacts to influence agent behavior during project manipulation.
Ingestion points: Methods such as portal.open_project, hmi.import_hmi_tags, plc.import_blocks, plc.import_cfc_charts, and project.import_cax_data across various reference files.
Boundary markers: The instructions emphasize mandatory 'explicit mutation authorization' and 'exact selectors' to maintain control, though they do not specify technical delimiters for the data content itself.
Capability inventory: Extensive capabilities including project lifecycle management (creation/deletion), file system writes (exports), and live industrial PLC operations (downloads and online state changes).
Sanitization: The skill recommends to 'Preflight source content' and 'Validate resulting types,' but lacks concrete implementation details for filtering or sanitizing embedded text within automation files.

tia-python