The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Gmail messages via gog gmail read and Google Sheets via gog sheets get which could contain malicious prompts.
Ingestion points: Email thread content and spreadsheet cell data are read into the agent context from SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the external data being processed to prevent the agent from following embedded instructions.
Capability inventory: The skill is authorized to update spreadsheets via gog sheets update and write to local files in the /Users/steipete/Projects/conferences/ directory.
Sanitization: There are no defined steps to sanitize or filter potential instructions embedded in the incoming communication data before it influences agent decisions or file writes.

speaking