The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @membranehq/cli package globally via npm. This is a vendor-owned resource for the Membrane platform and is necessary for the skill's functionality.
[COMMAND_EXECUTION]: The skill uses the membrane CLI to execute various shell commands for logging in, connecting to services, and listing or running actions. These are expected behaviors for a CLI-based integration.
[REMOTE_CODE_EXECUTION]: The skill utilizes the membrane action create command, which allows the Membrane platform to automatically generate actions based on user descriptions. These actions are then executed via membrane action run. This functionality is part of the managed Membrane platform's core service.
[PROMPT_INJECTION]: The skill ingests user-provided intent and parameters which are passed as arguments to the CLI commands. This represents an indirect prompt injection surface. \n * Ingestion points: User-provided strings for the --intent parameter in membrane action list, the description in membrane action create, and the JSON input in membrane action run.\n * Boundary markers: The use of structured JSON for action input parameters provides some separation between data and instructions.\n * Capability inventory: The skill can perform shipping and logistics operations (tracking, rating, labels, manifests) within the context of a Zenkraft connection.\n * Sanitization: Relies on the Membrane platform's internal processing of inputs received through its CLI.

zenkraft