The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local Node.js scripts (grid.cjs and verify.cjs) to handle complex diagram layouts and alignment checks. It instructs the agent to execute these scripts via the shell, which is a legitimate and scoped use of resources for the skill's primary purpose.
[PROMPT_INJECTION]: The skill demonstrates a surface for indirect prompt injection as it processes user-provided labels and descriptions to generate the character grid.
Ingestion points: Diagram labels and structure descriptions are provided by the user and processed via SKILL.md logic.
Boundary markers: The protocol does not explicitly define delimiters for user-supplied strings when they are interpolated into the JSON specifications for the CLI tool.
Capability inventory: Subprocess execution for running node commands is used across all diagram rendering and verification steps.
Sanitization: The included rendering utility (grid.cjs) uses JSON.parse to process incoming specifications, ensuring that the data is treated as structured input rather than executable commands.

ascii-diagram