The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @membranehq/cli package. This is the official command-line tool provided by the vendor for interacting with their platform.
[COMMAND_EXECUTION]: The instructions include various membrane CLI commands for logging in, connecting to services, and running actions. These are legitimate administrative and operational commands for the service.
[SAFE]: The skill encourages the use of managed connections and explicitly advises against asking users for API keys or secrets, which aligns with credential safety best practices.
[PROMPT_INJECTION]: The skill processes untrusted data from external Cloudmersive APIs (e.g., OCR text extraction, entity recognition) which represents an indirect prompt injection surface.
Ingestion points: Data returned from Cloudmersive actions such as extract-text-from-image or analyze-sentiment in SKILL.md.
Boundary markers: None identified in the instructional text.
Capability inventory: The skill facilitates command execution through the membrane action run command as described in SKILL.md.
Sanitization: No explicit sanitization or filtering of API outputs is mentioned in the provided instructions.

cloudmersive