The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @membranehq/cli package globally via npm. This is a legitimate vendor resource provided by the author to enable communication with the Membrane platform.
[COMMAND_EXECUTION]: The skill executes multiple shell commands using the membrane CLI to handle authentication, manage connections, and execute actions. These commands are necessary for the skill's primary function and use the vendor's official tooling.
[DATA_EXFILTRATION]: Sensitive credentials are not stored or requested by the skill; instead, it uses membrane login to handle authentication server-side, which is a secure method for managing access tokens without exposing them to the agent context.
[PROMPT_INJECTION]: The skill processes external data retrieved from Landbot, such as customer messages and profiles. This creates an indirect prompt injection surface where malicious data from a chatbot conversation could attempt to influence the agent's behavior. The agent should be configured to treat all data returned from Landbot actions as untrusted.

landbot