The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates user-supplied descriptions into generated SVG files without sanitization.
Ingestion points: User input via $ARGUMENTS is used to determine the content and text within the generated SVG.
Boundary markers: Absent. The skill does not instruct the agent to use delimiters or ignore instructions embedded within the user's description.
Capability inventory: The skill possesses Write capabilities to create SVG files and Bash capabilities to execute the open command.
Sanitization: Absent. There are no instructions for the agent to escape or validate user-provided strings before placing them in SVG <text> elements, which could allow for the injection of malicious markup or scripts if the SVG is viewed in a web browser.
[COMMAND_EXECUTION]: The workflow instructs the agent to use the open command via Bash to preview generated SVG files.
Evidence: The instruction Open with open <file.svg> for preview in the Step 2 workflow section.
While used for the legitimate purpose of previewing, this command acts on files whose content is partially derived from untrusted user input.

aris-pixel-art