sigma
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
membraneCLI to perform various tasks including user authentication (membrane login), searching for resources (membrane search), and executing API actions (membrane action run).- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the@membranehq/clipackage globally via npm, which is a required dependency for the skill's operations.- [PROMPT_INJECTION]: The skill processes data from the Sigma platform, creating a surface for indirect prompt injection where malicious content in the analytics data could attempt to influence the agent's behavior. - Ingestion points: Data retrieved through
membrane action runandmembrane requestcommands. - Boundary markers: None identified in the skill instructions.
- Capability inventory: CLI-based network requests and data manipulation commands.
- Sanitization: No sanitization or validation of the external data is specified before it is returned to the agent context.
Audit Metadata