The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external metadata, which is untrusted input, without explicit instruction boundaries or sanitization protocols.
Ingestion points: The skill is designed to "Process metadata extractor operations" and work with "metadata" input as described in SKILL.md.
Boundary markers: There are no defined delimiters or instructions to the agent to ignore embedded commands within the processed metadata.
Capability inventory: The skill allows access to powerful tools including Bash(cmd:*), Read, Write, and Edit, as listed in the allowed-tools section of SKILL.md.
Sanitization: No sanitization, validation, or escaping logic is described for the incoming metadata before it is processed by the agent or passed to system tools.

metadata-extractor