The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (scripts/score.py) to perform arithmetic for a decision matrix. The script is invoked via a shell command using numeric flags derived from the agent's evaluation of user input. The script includes basic validation to ensure inputs are within a 0-5 range.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze untrusted user-provided product ideas. There is an inherent risk that a maliciously crafted idea could include instructions designed to influence the agent's scoring, override the decision logic, or manipulate the final verdict.
[DATA_EXFILTRATION]: The skill features a file-writing capability where the agent saves results to a Markdown file in a directory named by the user. This creates an arbitrary file-write surface; without strict path validation, an agent could be manipulated into writing or overwriting files in sensitive directories if the underlying platform permissions allow it.

is-this-an-app