The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli package from the NPM registry. This package is the official tool provided by the vendor (Membrane) and is necessary for the skill to interact with the Nylas integration platform.
[COMMAND_EXECUTION]: The skill functions by executing various shell commands via the membrane CLI, such as membrane login, membrane connect, and membrane action run. These are standard operations for managing the integration.
[DATA_EXFILTRATION]: The skill implements secure authentication patterns. It utilizes the membrane login command to handle OAuth flows and token management, and it explicitly instructs the agent not to ask for or handle sensitive API keys directly, which prevents potential credential exposure.
[PROMPT_INJECTION]: The skill uses natural language inputs for searching and creating actions (e.g., membrane action list --intent "QUERY"). While this creates a surface for indirect prompt injection (Category 8), the risk is minimized as the actions are executed within the controlled environment of the Membrane platform.

nylas