zoho-cliq
Pass
Audited by Gen Agent Trust Hub on May 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
@membranehq/clipackage via NPM, which is a vendor-provided tool for managing integrations. - [COMMAND_EXECUTION]: The skill uses the
membraneCLI to perform operations such as logging in, creating connections, and running actions, which involves executing shell commands to interact with the Membrane platform. - [DATA_EXFILTRATION]: The skill facilitates network communication between the user's environment, Zoho Cliq, and the Membrane platform. This data flow is central to the skill's purpose, and credentials are managed server-side by the vendor to prevent local exposure.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8) because it retrieves and processes untrusted data from Zoho Cliq.
- Ingestion points: External data is ingested via the
membrane action runcommand in SKILL.md. - Boundary markers: No specific boundary markers or instructions are provided to the agent to distinguish between its instructions and the retrieved data content.
- Capability inventory: The skill can perform network operations and build new actions dynamically using the
membrane action createcommand. - Sanitization: No sanitization or validation logic is mentioned for the data fetched from Zoho Cliq APIs.
Audit Metadata