The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill ingests untrusted data from the user's Notion workspace, which could potentially contain malicious instructions aimed at influencing the synthesis process. While common for research tools, it warrants review during the synthesis step.
Ingestion points: The skill retrieves content using the Notion:notion-fetch tool as defined in SKILL.md.
Boundary markers: The skill does not explicitly define sanitization or boundary markers in its instructions, though it encourages the use of structured templates (reference/format-selection-guide.md) and citations (reference/citations.md).
Capability inventory: The skill possesses write capabilities through Notion:notion-create-pages and Notion:notion-update-page as specified in SKILL.md.
Sanitization: No explicit sanitization of external content is mentioned before it is processed by the AI.
[External Service Reference]: The skill provides instructions for connecting to the official Notion Model Context Protocol (MCP) server. This is a standard setup procedure for integrating with the Notion platform.
Evidence: SKILL.md and agents/openai.yaml reference https://mcp.notion.com/mcp for the Notion MCP server.

notion-research-documentation