The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes instructions to run which runwb2, which ansys, and python -c imports to verify the environment's capabilities in references/software-notes.md. These are standard diagnostic probes used to identify available tools.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow in references/community-troubleshooting.md for searching external community forums and GitHub issues. Ingesting content from these untrusted external sources represents an attack surface where malicious instructions could be embedded in data processed by the agent.
Ingestion points: Search results from ANSYS Learning Forum, GitHub Issues, and Stack Overflow (referenced in references/community-troubleshooting.md).
Boundary markers: The skill includes instructional boundary markers, telling the agent to "Reject license bypasses, unsafe downloads, and mismatched product-family advice."
Capability inventory: The skill can perform environment checks via shell commands and generate scripts for ANSYS solvers (references/software-notes.md).
Sanitization: No technical sanitization or filtering of external content is specified beyond the instructional warning.

ansys-agent