The Agent Skills Directory

[COMMAND_EXECUTION]: The procedure in SKILL.md (Step 1.3) instructs the agent to execute a shell command: python3 scripts/validate-metadata.py --name "[name]" --description "[description]". This pattern of using raw placeholders for shell arguments is a potential command injection vector.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the ingestion of untrusted user data used in file creation and command execution.
Ingestion points: User-supplied values for name and description in SKILL.md Step 1.
Boundary markers: None present to delimit user input from the rest of the command string.
Capability inventory: The skill performs shell command execution (python3) and file system modifications (directory and file creation in Steps 2 and 3).
Sanitization: None. While a validation script is provided, it is intended to be executed with the untrusted input rather than sanitizing it beforehand.
[SAFE]: The script scripts/validate-metadata.py is incomplete. The provided code defines a validation function but fails to implement argument parsing or a function call in the if __name__ == "__main__": block, making it non-functional.

skill-creator