The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the terminal for running a provided Python script (scripts/upload.py) to upload diagram files to a remote service.
[EXTERNAL_DOWNLOADS]: The documentation identifies the cryptography Python package as a dependency for the upload script, which is a reputable and standard library.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured JSON data, creating a potential surface for indirect prompt injection. Ingestion points: Data enters the context through diagram element arrays and user-provided JSON content in SKILL.md. Boundary markers: No specific delimiters are mandated to differentiate instructions from data content. Capability inventory: The skill requires access to file system writing and terminal execution for its core features. Sanitization: There is no mention of input validation or sanitization for the generated JSON elements.

excalidraw