The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local metadata files and external keyword analysis tools, which constitutes a potential attack surface for indirect prompt injection.
Ingestion points: The skill reads local files at metadata/app-info/{locale}.json and metadata/version/{latest-version}/{locale}.json. It also ingests data from external competitors via the extract_competitors_keywords tool in Astro MCP.
Boundary markers: The instructions do not specify any delimiters or safety boundaries for the data being read, nor do they include warnings to the agent to ignore instructions embedded within the metadata.
Capability inventory: The skill has access to the asc CLI tool for managing App Store credentials (asc auth), search plans, and metadata synchronization. It also uses Astro MCP for keyword data operations.
Sanitization: There is no evidence of content validation or sanitization for the metadata fields (description, keywords, subtitle) before they are processed by the agent.
[COMMAND_EXECUTION]: The skill utilizes the asc CLI tool to perform various App Store management tasks.
The skill instructs the agent to execute shell commands such as asc app-tags list, asc optimize search plan, and asc metadata keywords apply. While these are legitimate functions for an ASO audit tool, they involve the execution of system commands based on skill logic and parameters.

asc-aso-audit