The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local scripts (scripts/canvas_template.py and scripts/canvas_validate.py) via the python3 interpreter. User-provided data, including the canvas title and various template-specific parameters, is interpolated directly into the shell command strings.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection (Category 8) where malicious instructions could be embedded in the user-provided parameters.
Ingestion points: User-provided title and template parameters (e.g., slide_count, step_count) defined in SKILL.md.
Boundary markers: The instructions attempt to use double quotes as delimiters (e.g., title="[title]") to wrap user input in the shell command.
Capability inventory: The skill has the capability to execute shell commands and write to the local filesystem at the specified output_path as seen in SKILL.md.
Sanitization: There is no evidence in the provided SKILL.md of input validation or sanitization to prevent characters like backticks, semicolons, or additional quotes from breaking out of the command string.

canvas-template