The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs multiple shell executions of local Python scripts (scripts/prompt-ab-tester.py and scripts/eval-runner.py) to manage tests and evaluate content. These executions include several user-controlled variables.
[PROMPT_INJECTION]: The skill is vulnerable to indirect command injection through the interpolation of user-supplied fields such as {content_or_path}, {name}, {label}, and {description} directly into shell command strings. An attacker could potentially supply malformed input (e.g., containing shell metacharacters like semicolons or backticks) to execute unauthorized commands on the host system.
Ingestion points: Untrusted data enters the context via the 'Content for the variant' and 'Test name' inputs.
Boundary markers: There are no explicit boundary markers or instructions to escape special characters before command execution.
Capability inventory: The skill has the capability to run subprocesses and read files from the user's home directory.
Sanitization: The instructions do not specify any sanitization or validation of the input before it is passed to the shell.
[COMMAND_EXECUTION]: The skill accesses structured data and configuration files located in the user's home directory (~/.claude-marketing/brands/), which, while necessary for its stated function, increases the impact of any successful command injection.

prompt-test