notion

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and processing untrusted data from Notion.
  • Ingestion points: Data enters the agent's context through page retrieval and database query endpoints specified in SKILL.md.
  • Boundary markers: The instructions do not define delimiters or explicit warnings to treat retrieved Notion data as untrusted.
  • Capability inventory: The skill allows the agent to perform various Notion operations using curl commands, including read, write, and search.
  • Sanitization: No explicit sanitization or validation of the retrieved content is mentioned before it is processed by the agent.
  • [COMMAND_EXECUTION]: The skill uses shell commands for environment configuration and API interaction.
  • Evidence: Shell commands mkdir, echo, and cat are used to initialize and read the API key from the ~/.config/notion/ directory.
  • Evidence: The tool uses curl to execute HTTP requests to the official Notion API domain api.notion.com.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 12:15 AM