The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the vendor's command-line tool (@membranehq/cli) globally via NPM to handle integration logic and authentication.
[COMMAND_EXECUTION]: The skill utilizes the membrane CLI to execute various operations, including authentication, connection management, and the execution of pre-built or dynamically created actions.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface due to its core functionality of ingesting and processing data from external web sources via Exa.
Ingestion points: Data returned from membrane action run (such as search results and web page content) enters the agent's context (SKILL.md).
Boundary markers: Absent; the instructions do not provide delimiters or warnings to treat external data as untrusted.
Capability inventory: Includes shell command execution, dynamic action creation, and network-enabled actions via the Membrane platform.
Sanitization: Absent; content fetched from external sources is processed without validation or filtering.

exa