The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data fetched from Notion pages and databases, which serves as a vector for indirect prompt injection. Maliciously crafted content within a Notion page could attempt to hijack the agent's instructions during the synthesis and document creation process.
Ingestion points: Data is ingested via Notion:notion-fetch and Notion:notion-query-data-sources as described in SKILL.md and the evaluations/ directory.
Capability inventory: The skill possesses write capabilities including Notion:notion-create-pages and Notion:notion-create-comment to generate meeting materials and notifications.
Boundary markers: The instructions do not define clear delimiters or provide 'ignore' directives to the agent regarding instructions potentially contained within the fetched Notion data.
Sanitization: There is no evidence of content sanitization, filtering, or validation performed on the retrieved Notion data before it is interpolated into new meeting documents.

notion-meeting-intelligence