The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the membrane CLI for operations including login, connection setup, and action execution. These commands are integral to the skill's purpose and use the vendor's specialized tooling.
[EXTERNAL_DOWNLOADS]: The instructions guide users to install the @membranehq/cli package globally via npm. This package is hosted on the public npm registry and is the official tool for the Membrane platform.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8) by using natural language strings for action discovery and creation without explicit boundary markers or sanitization within the command structure.
Ingestion points: Natural language input used in CLI arguments like --intent and --description.
Boundary markers: Not present in the provided command templates.
Capability inventory: The skill is capable of running logic against a connected service and creating new actions dynamically.
Sanitization: No input validation or filtering is defined in the skill's instructions.

aha