The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @membranehq/cli package globally via npm. This is a vendor-owned package required for the skill to function as intended.
[COMMAND_EXECUTION]: The skill makes extensive use of the membrane CLI to perform operations such as logging in, connecting to services, listing actions, and running those actions. These commands interact with both the local environment and the Membrane cloud platform.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it retrieves and processes data from Amazon Elasticsearch instances.
Ingestion points: Data ingested from Elasticsearch via membrane action run or metadata from membrane action list is processed by the agent.
Boundary markers: The instructions do not define clear delimiters or include warnings for the agent to ignore potentially malicious instructions embedded in the data retrieved from the database.
Capability inventory: The agent has the ability to execute further actions via membrane action run and create new actions via membrane action create based on its interpretation of the data.
Sanitization: There is no explicit sanitization or validation logic described to filter or escape content retrieved from the external data source.

amazon-elasticsearch