The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the official @postman/postman-mcp-server package via NPM and configures the agent to communicate with Postman's own remote servers at mcp.postman.com and mcp.eu.postman.com. These are recognized well-known services.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect injection as it processes untrusted external data.
Ingestion points: Tools like create_api_spec ingest schema content (OpenAPI), and the agent reads collection/environment data from the Postman API.
Boundary markers: No explicit boundary markers or instructions to ignore embedded prompts are provided in the skill documentation.
Capability inventory: The skill has capabilities for network operations (send_request, run_collection) and code generation (generate_client_code), which could be targets for manipulation if a malicious spec is processed.
Sanitization: No specific sanitization or validation steps for external schema content are described in the setup instructions.
[OBFUSCATION]: A Base64-encoded configuration string is utilized within the Cursor installation URL to pass setup parameters.
Evidence: The config parameter in the cursor.com/en/install-mcp link decodes to a standard JSON object containing the server URL and a placeholder for the API key.

postman-mcp-server