notion-knowledge-capture

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions to set up the Notion Model Context Protocol (MCP) using the official URL https://mcp.notion.com/mcp. This is a well-known and trusted service.
  • [COMMAND_EXECUTION]: Includes setup steps that guide the user to configure the agent's environment using the codex CLI (e.g., codex --enable rmcp_client and codex mcp login). These are standard configuration commands for the tool's intended functionality.
  • [PROMPT_INJECTION]: The skill is designed to process untrusted text (conversations) which presents an inherent surface for indirect prompt injection.
  • Ingestion points: Conversation history and user-provided notes are ingested during the extraction phase (SKILL.md).
  • Boundary markers: Absent; the instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) to isolate user-provided data from agent instructions.
  • Capability inventory: The skill has the capability to search, fetch, create, and update pages in the user's Notion workspace (Notion:notion-search, Notion:notion-fetch, Notion:notion-create-pages, Notion:notion-update-page).
  • Sanitization: Content is transformed into structured markdown for Notion; however, no explicit sanitization or instruction-filtering logic is defined in the workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 08:24 PM