The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, specifically mobile application files (APKs, IPAs, and source code), which creates a surface for indirect prompt injection.
Ingestion points: The skill utilizes codebase and filesystem tools to read mobile application assets and source files.
Boundary markers: The instructions do not include specific delimiters or warnings to treat content from analyzed files as untrusted data, increasing the risk that the agent might follow malicious instructions embedded within app metadata or code comments.
Capability inventory: The agent has access to the local filesystem and the codebase of the project it is auditing.
Sanitization: There is no mention of sanitizing or validating the content extracted from mobile applications before processing it.

mobile-security-tester