The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose is to analyze untrusted external text (papers, essays, and documents) which may contain malicious instructions designed to subvert the agent's detection logic.
Ingestion points: Untrusted data enters the agent's context through the reading of documents targeted for analysis in SKILL.md.
Boundary markers: The skill does not define clear delimiters or specific instructions for the agent to ignore potential commands embedded within the analyzed text.
Capability inventory: The skill has access to WebSearch and WebFetch tools to perform citation verification; while these tools are used for validation, they represent a capability surface if an injection were successful.
Sanitization: There is no evidence of input sanitization or filtering to prevent the agent from interpreting content within the analyzed documents as instructions.

ai-detect