The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions guide the agent to execute shell commands using the cadgen tool to manage the viewer lifecycle.
Evidence: Commands such as cadgen viewer, cadgen doctor, cadgen viewer list, and cadgen viewer stop are used for starting, verifying, and stopping the visualization server.
[EXTERNAL_DOWNLOADS]: The skill specifies a dependency on an external Python package from the vendor.
Evidence: The requirements.txt file includes cadgen==0.6.3, which is downloaded and installed to provide the necessary server and client components.
[INDIRECT_PROMPT_INJECTION]: The skill processes external CAD and robot-description files, which serves as a potential surface for indirect prompt injection.
Ingestion points: The agent is instructed to read and resolve paths for various file types such as .step, .stp, .glb, .stl, .3mf, .dxf, .urdf, .srdf, and .sdf (as documented in SKILL.md and references/viewer-features.md).
Boundary markers: There are no explicit instructions for the agent to use delimiters or to disregard potential instructions embedded within the CAD data during processing.
Capability inventory: The skill allows for the execution of CLI commands and the initiation of a local network server.
Sanitization: The documentation does not specify any sanitization or validation protocols for the input files before they are handled by the viewer.

cad-viewer