The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md and README.md (Step 2) direct the agent to execute a shell command (node "$OD_BIN" media generate ...) to produce image assets. This command uses environment variables ($OD_BIN, $OD_PROJECT_ID) and parameters derived from user input.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to command injection because it interpolates user-controlled data—specifically the <short-descriptive-name> and the <full assembled prompt>—directly into a shell command string without prescribing any sanitization, escaping, or validation.
Ingestion points: User-provided briefs and metadata (model, aspect, style) are used to construct the command arguments in SKILL.md and README.md.
Boundary markers: The instructions lack delimiters or warnings to treat the interpolated user content as untrusted data within the shell context.
Capability inventory: The skill requires the agent to have shell execution capabilities to run the image generation tool.
Sanitization: There is no instruction to sanitize shell metacharacters (such as ;, &, |, or backticks) that could be used to break out of the intended command and execute arbitrary code.

image-poster