The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the execution of internal Python scripts (scripts/step, scripts/inspect, scripts/snapshot) to generate, validate, and document CAD models. These tools are invoked via the CLI and operate on files within the project workspace.
[EXTERNAL_DOWNLOADS]: The skill requires the playwright package, as specified in requirements.txt, which is used for headless browser rendering of CAD geometry snapshots. This is a standard dependency for visual automation tasks.
[DYNAMIC_EXECUTION]: The core functionality involves generating and executing Python code that utilizes the build123d library. The agent authors these scripts based on user requirements and executes them to produce STEP artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data such as natural language specifications, reference images, and technical drawings. It mitigates risks by implementing a 'CAD brief' workflow that requires the agent to extract and validate dimensions and intent before code generation.
[SAFE]: Analysis of the snapshot rendering logic in scripts/snapshot/__main__.py reveals security best practices, including path validation and origin checks to prevent directory traversal when serving local assets to the browser environment.

cad