notion

Warn

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [CREDENTIALS_UNSAFE] (MEDIUM): The skill recommends storing a sensitive API key in a plain text file at ~/.config/notion/api_key. Unencrypted storage of credentials on the local filesystem poses a risk of exposure.
  • [DATA_EXFILTRATION] (MEDIUM): The skill reads the ~/.config/notion/api_key file and sends its content to api.notion.com via an Authorization header. While functionally necessary, this behavior involves accessing sensitive user files and transmitting data to an external endpoint.
  • [Indirect Prompt Injection] (LOW): The skill ingests external data from the Notion API, which could include adversarial instructions intended to manipulate agent behavior.
  • Ingestion points: Results from curl calls to search, page, and block endpoints in SKILL.md.
  • Boundary markers: No boundary markers or isolation techniques are implemented.
  • Capability inventory: The skill uses curl for API operations; no dynamic execution of retrieved content is detected.
  • Sanitization: No sanitization of ingested content from Notion is documented.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 18, 2026, 11:59 AM