cloudmersive
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Installs the official Membrane CLI package (@membranehq/cli) from the npm registry, which is the intended tool for the platform.
- [COMMAND_EXECUTION]: Executes commands via the membrane CLI to interact with the Cloudmersive API. This method is used to manage connections and run actions without exposing raw API keys.
- [PROMPT_INJECTION]: Exposure to indirect prompt injection through processing untrusted data from API outputs, such as OCR text or website scan results.
- Ingestion points: Data returned from
membrane action runandmembrane requestcommands in SKILL.md. - Boundary markers: No specific delimiters or warnings are used to isolate untrusted data in the prompt templates.
- Capability inventory: Includes shell command execution and network requests via the
membranetool across all interaction examples. - Sanitization: The instructions do not include explicit sanitization or validation of the external data before it is presented to the agent.
Audit Metadata