gtm-run
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external website URL, which represents a potential surface for indirect prompt injection attacks.
- Ingestion points: The skill ingests a website URL via the
$ARGUMENTSparameter as defined inSKILL.md. - Boundary markers: The orchestration logic in
SKILL.mddoes not specify explicit delimiters or "ignore embedded instructions" warnings for the content fetched from the external URL. - Capability inventory: The skill possesses capabilities to search Notion, create/modify Notion pages, and chain multiple specialized sub-skills (
gtm-strategy:*). - Sanitization: No specific sanitization or escaping mechanisms for the external web content are described in the orchestrator file.
- Risk Mitigation: The skill includes four mandatory human-in-the-loop checkpoints (GATE 1, 2, 2.5, and 3), requiring explicit user approval before the agent proceeds to deeper levels of analysis or action, significantly reducing the impact of potential injection.
Audit Metadata