The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands that incorporate unvalidated user input. Specifically, the <app> identifier provided by the user is passed directly to local scripts via python3 scripts/aso_optimize_v2.py analyze <app> and python3 scripts/aso_optimize_v2.py validate <app>. If the user provides an identifier containing shell metacharacters (such as ;, &&, or |), it could lead to the execution of arbitrary commands on the host system.
[PROMPT_INJECTION]: The workflow involves an indirect prompt injection surface where the agent ingests and processes data from external sources. In Step 2, the agent reads app metadata (titles, subtitles, and keywords) fetched from the App Store to perform tokenization. This data is external and potentially attacker-controlled. The skill lacks boundary markers or specific instructions to sanitize this content or to treat it as untrusted data, which could allow malicious instructions embedded in app metadata to influence the agent's behavior.

aso-optimize