NYC

notion-research-documentation

Fail

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: HIGHPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (HIGH): The skill lacks isolation between instructions and the data it processes from Notion. 1. Ingestion points: Content enters via Notion:notion-search and Notion:notion-fetch (SKILL.md). 2. Boundary markers: No delimiters are used to wrap search results. 3. Capability inventory: The agent can write to the workspace via Notion:notion-create-pages. 4. Sanitization: No filtering of source content is performed.
  • [DATA_EXFILTRATION] (MEDIUM): An attacker could manipulate the research workflow to discover and centralize sensitive workspace data into new pages or databases.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 15, 2026, 11:09 PM