qonto-crew-onboard
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data provided by users (such as names, roles, and start dates) to drive automated actions like sending organization invitations and initiating card requests.
- Ingestion points: Data enters the workflow via user prompts such as "Alex starts Monday as a developer" or "onboard Sam as an account exec" in SKILL.md.
- Boundary markers: There are no explicit technical delimiters or markers defined to isolate user-provided text from internal instructions.
- Capability inventory: The skill has access to sensitive write-capable tools including
create_membership,create_team,create_card_request, andchange_card_statusin SKILL.md. - Sanitization: The skill mandates a "recap table" and individual confirmation for every write action, which serves as a human-in-the-loop sanitization mechanism to prevent unintended execution from malicious input.
Audit Metadata