The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and analyze external IFC files. As these files are untrusted data sources, they represent a surface for indirect prompt injection attacks where malicious instructions could be embedded within the file's metadata, property sets, or object descriptions.
Ingestion points: IFC files processed through the validation trees described in SKILL.md and references/methods.md.
Boundary markers: The skill does not define specific delimiters or instructions to ensure the agent ignores potential instructions embedded within the IFC data structures.
Capability inventory: The skill acts as an orchestrator, routing findings to other ifc-errors-* skills. It does not contain its own scripts for file modification or network exfiltration, though it provides instructional context for command-line tools.
Sanitization: No explicit sanitization or validation of the text content within IFC attributes is provided to mitigate injection risks.
[EXTERNAL_DOWNLOADS]: The skill references official resources from buildingSMART, including the validation service at validate.buildingsmart.org and the normative Gherkin rules on GitHub. These are well-known, authoritative sources for the domain and are used appropriately for documentation and reference.
[COMMAND_EXECUTION]: Documentation within references/methods.md provides an example command for debugging Gherkin rules locally (python test/test_main.py). This is provided for developer guidance and is not an automated execution path within the skill's operational logic.

ifc-agents-file-validator