The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the @membranehq/cli package, which is the official tool provided by the vendor for platform interaction.
[COMMAND_EXECUTION]: The skill uses the membrane command-line interface to perform authentication, connector discovery, and API actions. These commands are necessary for the skill's documented functionality.
[DATA_EXFILTRATION]: All API communication is routed through a managed proxy service provided by the vendor. This architecture allows the platform to handle sensitive credentials and token refreshes server-side, preventing the exposure of API keys in the agent's local environment.
[PROMPT_INJECTION]: This skill acts as a data bridge for Benchmark Email records. Evidence chain: 1. Ingestion points: Records are fetched using membrane action run and membrane request in SKILL.md. 2. Boundary markers: The skill does not provide explicit markers to separate data from instructions. 3. Capability inventory: The skill utilizes the membrane CLI for network operations as described in SKILL.md. 4. Sanitization: No explicit sanitization of external data is performed. However, the use of structured JSON output and vendor-managed actions minimizes the risk of indirect injection.

benchmark-email