The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of raw interview notes and transcripts, which creates a surface for indirect prompt injection where malicious instructions embedded in the data could influence the agent's output.
Ingestion points: The skill ingests raw notes, transcripts, and external file references (@file.md) as specified in the 'Instructions' and 'Mode Behaviors' sections.
Boundary markers: The skill lacks explicit delimiters or instructions to the model to ignore embedded commands within the interview text being synthesized.
Capability inventory: The skill has permissions to search local project directories and write new markdown files containing the synthesized research to the filesystem.
Sanitization: There are no defined steps for sanitizing, filtering, or validating the content of the interviews before they are processed by the agent.

interview-synthesis