The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of interview transcripts provided by the user via file attachments or direct text input.
Ingestion points: The skill instructions in SKILL.md direct the agent to read attached files (text, PDF, audio transcription) or pasted transcript content.
Boundary markers: The instructions do not include explicit delimiters or guidance for the agent to ignore instructions that might be embedded within the interview transcript.
Capability inventory: The skill generates a markdown summary and saves it to the user's workspace. It does not utilize network operations, shell commands, or access to sensitive system files.
Sanitization: No input validation or sanitization mechanisms are present to filter content from the processed transcripts.

summarize-interview