The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the drawio command-line interface to export diagrams to PNG, SVG, and PDF formats. It also uses environment-specific commands (open, xdg-open, start) to open the generated files locally.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided articles to generate multiple technical diagrams. It lacks explicit boundary markers or instructions to the agent to ignore potentially malicious instructions embedded within the untrusted article text. This creates an attack surface where maliciously crafted input could attempt to influence the file naming or the resulting shell commands.
Ingestion points: User-provided articles and text descriptions (referenced in SKILL.md and references/use-cases.md).
Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore instructions within the source data.
Capability inventory: Shell command execution via drawio CLI and file system writes for .drawio and exported image files (detailed in references/export-and-files.md).
Sanitization: Absent. While XML entity escaping is mentioned for the diagram content, there is no explicit instruction to sanitize user-derived strings before they are used as filenames or shell arguments.

drawio-chart