duffel
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses the Membrane CLI for all operations, which is the official tool from the skill author.
- [SAFE]: Authentication is managed via an OAuth flow (membrane login), ensuring that no sensitive credentials or API keys are stored within the skill instructions.
- [SAFE]: The skill follows security best practices by instructing the agent to never ask for user secrets and to use server-side credential management.
- [SAFE]: The skill has a surface for indirect prompt injection as it processes external API data from Duffel; however, it relies on the vendor's platform for secure communication and does not exhibit dangerous autonomous behavior. 1. Ingestion points: Output from membrane action run and membrane request. 2. Boundary markers: Absent. 3. Capability inventory: Subprocess calls via membrane CLI. 4. Sanitization: Not explicitly implemented in prompt logic.
Audit Metadata