The Agent Skills Directory

[SAFE]: The skill instructions utilize the official @membranehq/cli package, which is the expected tool for the 'membranedev' vendor. Global installation of this CLI is a standard procedure for this platform.
[SAFE]: Authentication is performed using the membrane login and membrane connect commands. This approach delegates credential management to the platform's secure infrastructure rather than requesting or storing API keys directly, which is a significant security best practice.
[SAFE]: The skill does not contain any evidence of prompt injection, obfuscation, or unauthorized data exfiltration. All network operations and tool invocations are consistent with the integration's described purpose.

referralhero