The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands using the membrane CLI tool to manage logins, establish connections to Render, and run specific actions.
[EXTERNAL_DOWNLOADS]: The instructions require the global installation of the @membranehq/cli package from the npm registry. This is a legitimate vendor resource provided by the skill author.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface (Category 8) as it processes untrusted user data—specifically natural language 'intents', action 'descriptions', and JSON input payloads—which are then interpolated into CLI commands and executed as subprocesses.
Ingestion points: SKILL.md (via --intent, --input, and action description parameters).
Boundary markers: None identified in the provided instructions.
Capability inventory: The skill has the capability to install packages, perform network authentication, and execute arbitrary platform actions via the shell.
Sanitization: No explicit sanitization or validation of user input is described in the skill's instructions.
[DYNAMIC_EXECUTION]: The skill utilizes membrane action create to generate new executable actions at runtime based on provided descriptions, which are subsequently polled and executed.

render