setup-webhook
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Provides instructions for the user to manually add a remote MCP server using npx.
- Evidence:
claude mcp add vapi-docs -- npx -y mcp-remote https://docs.vapi.ai/_mcp/serverinSKILL.md. - Context: The command targets an official vendor-controlled documentation endpoint.
- [COMMAND_EXECUTION]: Includes standard shell commands for API interaction and local development testing.
- Evidence: Uses
curlfor PATCH requests toapi.vapi.aiand providesvapi listencommands for terminal-based event forwarding inSKILL.md. - [INDIRECT_PROMPT_INJECTION]: Identifies an ingestion surface where the assistant processes data from external webhook events.
- Ingestion points: The
/vapi/webhookPOST endpoint defined in the provided Express (TypeScript) and Flask (Python) code examples inSKILL.md. - Boundary markers: Absent in the simplified code examples.
- Capability inventory: The provided examples are limited to logging incoming messages and returning JSON assistant configurations; no dangerous file system or shell operations are present.
- Sanitization: The examples utilize standard JSON parsing of the request body.
Audit Metadata