The Agent Skills Directory

[SAFE]: The skill uses the official Membrane CLI (@membranehq/cli) for all operations. This is a vendor-provided tool that handles authentication securely via OAuth flows.
[EXTERNAL_DOWNLOADS]: Installs the @membranehq/cli package via npm. This is a standard dependency for skills interacting with the Membrane platform and originates from the vendor's official package registry.
[CREDENTIALS_UNSAFE]: The skill explicitly instructs against hardcoding API keys or tokens. It uses a delegated connection model (membrane connect) where secrets are managed server-side by the platform, minimizing the risk of local credential exposure.
[COMMAND_EXECUTION]: Shell commands are restricted to the membrane CLI for action discovery, execution, and proxy requests, which is the intended and documented purpose of the skill.

clay