The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the membrane CLI for searching connectors, managing connections, and executing API actions. This centralized command usage allows the platform to manage security boundaries through a single interface.
[EXTERNAL_DOWNLOADS]: The skill directs the user or agent to install the @membranehq/cli package from the NPM registry and optionally run it via npx. These are official distribution channels for the vendor's tooling and are considered safe.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and process untrusted data (articles, comments, and user profiles) from InSided communities.
Ingestion points: Data retrieved via membrane action run and membrane request as described in SKILL.md.
Boundary markers: None are specified in the current skill instructions to isolate retrieved content from the agent's control flow.
Capability inventory: The skill utilizes membrane CLI commands for network communication and data retrieval (SKILL.md).
Sanitization: The instructions do not currently include specific sanitization or filtering steps for the retrieved content.

insided