The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli package via NPM.
Evidence: npm install -g @membranehq/cli@latest in SKILL.md.
This is a vendor-owned resource used for the skill's core functionality.
[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands to interact with the Membrane platform.
Evidence: Commands such as membrane login, membrane connect, and membrane action run are used throughout SKILL.md.
These commands are necessary for the skill's purpose and are used to manage connections and execute API actions.
[PROMPT_INJECTION]: The skill ingests data from external PassSlot actions which could contain instructions intended to influence agent behavior (Indirect Prompt Injection).
Ingestion points: Data returned from membrane action run commands is processed by the agent.
Boundary markers: No specific delimiters or boundary markers are defined to isolate external data from instructions.
Capability inventory: The skill utilizes the membrane CLI to perform operations (SKILL.md).
Sanitization: No sanitization of the external action output is described before the data is processed by the agent.
[CREDENTIALS_UNSAFE]: The skill follows security best practices by using a platform-based authentication flow.
Evidence: It explicitly instructs: "Let Membrane handle credentials — never ask the user for API keys or tokens."
This prevents the exposure of sensitive credentials within the local environment or conversation logs.

passslot