The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @membranehq/cli package from the public npm registry. This is a legitimate utility provided by the skill's author to enable interaction with the Membrane platform.
[COMMAND_EXECUTION]: The skill relies on executing shell commands via the membrane CLI to perform core functions such as logging in, connecting to services, and running data actions.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by incorporating user-controlled input into CLI command arguments.
Ingestion points: User-provided strings are interpolated into the --intent parameter of membrane action list and the --input parameter of membrane action run in SKILL.md.
Boundary markers: No delimiters or protective warnings are used to isolate user data from the command structure.
Capability inventory: The skill can execute and create remote actions using the membrane action run and membrane action create commands documented in SKILL.md.
Sanitization: The instructions do not define any sanitization or validation steps for external content before processing.
[SAFE]: The skill follows security best practices by delegating authentication to the Membrane platform via a browser-based flow, ensuring that API keys and tokens are managed securely and not stored in plain text by the agent.

rex