The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides code examples that ingest untrusted data from external sources, creating a surface for indirect prompt injection.
Ingestion points: references/examples.md and references/methods.md demonstrate opening IFC files using ifcopenshell.open().
Boundary markers: None are used in the examples to delimit or ignore instructions within the loaded IFC data.
Capability inventory: The skill provides instructions for ifcopenshell.api.run (modifying model state) and model.write (writing files to the system) across multiple files.
Sanitization: No sanitization or validation of the ingested IFC content is present in the provided snippets.
[DATA_EXFILTRATION]: No network activity or attempts to access sensitive system files (e.g., credentials, SSH keys) were detected. File operations are restricted to standard BIM model outputs.
[REMOTE_CODE_EXECUTION]: No patterns for downloading or executing remote code or scripts from untrusted URLs were found.
[COMMAND_EXECUTION]: Command execution is limited to standard Python and Blender API calls required for BIM modeling.
[SAFE]: The skill's metadata and instructional content are consistent with the legitimate OpenAEC-Foundation Bonsai project and represent standard industry practices for OpenBIM development.

bonsai-syntax-spatial