The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted conversation data to generate Notion content, creating a surface for indirect prompt injection.
Ingestion points: The workflow in SKILL.md (Step 1 and 3) directs the agent to extract information from potentially untrusted conversation history.
Boundary markers: There are no specific delimiters or "ignore embedded instructions" warnings defined to isolate processed text from agent instructions.
Capability inventory: The agent uses tools like Notion:notion-create-pages and Notion:notion-update-page to write content to external Notion databases.
Sanitization: The instructions lack guidance for escaping or validating content extracted from conversations before it is passed to the Notion tools.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the official Notion MCP from a remote URL.
Evidence: SKILL.md includes the command codex mcp add notion --url https://mcp.notion.com/mcp.
Context: The URL belongs to a well-known service and the instruction is part of a standard configuration workflow.

notion-knowledge-capture