The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run() within the IFCExporter class to execute the IfcExporter.exe binary. While the implementation uses a list-based argument structure to mitigate shell injection, the execution of external binaries is a high-privilege operation.
[EXTERNAL_DOWNLOADS]: The documentation and metadata reference external dependencies including IfcExporter, IfcConvert, and IfcOpenShell. These are external tools required for the skill's functionality, sourced from the vendor's website (datadrivenconstruction.io) or community repositories.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection (Category 8).
Ingestion points: Data is ingested from external IFC files via the IfcExporter and subsequently read from Excel sheets using pandas.read_excel (e.g., in read_elements and get_properties).
Boundary markers: No explicit boundary markers or 'ignore' instructions are present when the agent processes the extracted BIM metadata.
Capability inventory: The skill has filesystem permissions and the ability to execute subprocesses via subprocess.run.
Sanitization: There is no evidence of sanitization or filtering for the metadata extracted from the IFC files before it is presented to the agent for analysis.
[DATA_EXFILTRATION]: The skill requests filesystem permissions to perform its primary function of reading model files and writing reports. While no network exfiltration was detected, the access level allows reading potentially sensitive BIM data from the local environment.

ifc-to-excel