The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli package from the NPM registry. This is the official tool provided by the vendor to interface with the Membrane platform.
[COMMAND_EXECUTION]: The agent is instructed to use the membrane CLI locally to perform authentication, list active connections, and execute actions.
[REMOTE_CODE_EXECUTION]: Through the action create command, the skill triggers the generation of logic on the vendor's remote platform based on natural language descriptions. This generated logic is then executed via the action run command.
[PROMPT_INJECTION]: The skill processes natural language input to discover or create actions, creating a surface for indirect prompt injection.
Ingestion points: Natural language strings are ingested via the --intent flag in action list and the DESCRIPTION argument in action create (found in SKILL.md).
Boundary markers: There are no specified delimiters or boundary instructions provided to help the agent isolate potentially untrusted input from its core instructions.
Capability inventory: The skill is capable of executing shell commands (CLI) and triggering the creation and execution of remote logic.
Sanitization: The instructions do not provide for sanitization or validation of the input strings before they are passed to the CLI commands.

datawrapper