The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Membrane CLI (membrane) to manage authentication (membrane login), established connections (membrane connect), and the execution of Firebase-related actions (membrane action run). These are standard operational commands for the vendor's platform.
[EXTERNAL_DOWNLOADS]: The skill documentation guides the user to install the official CLI tool @membranehq/cli from the NPM registry. This package is managed by the skill author and is necessary for the intended functionality.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by interpolating natural language descriptions into action creation and search queries.
Ingestion points: User-provided strings for the --intent and --description parameters in SKILL.md.
Boundary markers: Not present in the command-line examples provided in the instructions.
Capability inventory: The skill allows the execution of generated actions using membrane action run as described in SKILL.md.
Sanitization: The skill relies on the backend Membrane platform to sanitize and validate the synthesis of actions from natural language input.
[CREDENTIALS_UNSAFE]: The skill correctly follows security best practices by advising against requesting API keys or tokens directly from the user, instead delegating credential management to the Membrane platform.

firebase-admin-sdk