The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (such as project names, aliases, and IP CIDR blocks) to configure cloud infrastructure.
Ingestion points: User requests in chat are used to populate request bodies for project creation, updates, and traffic filter rules.
Boundary markers: The skill lacks explicit instructions for the agent to use delimiters when interpolating user data into CLI inputs, though it utilizes JSON files for command parameters which provides some structural separation.
Capability inventory: The skill executes elastic CLI commands to create, update, and delete Serverless projects and Hosted deployments, and to modify network traffic filters.
Sanitization: There are no specific instructions for the agent to sanitize or validate the length/content of user-supplied strings before passing them to the CLI.
[COMMAND_EXECUTION]: The skill relies on the elastic CLI (>= 0.2) to perform all control-plane operations.
Evidence: Operations are mapped to commands such as elastic cloud serverless projects create and elastic cloud hosted deployments update-deployment.
Context: This is legitimate functionality for the skill's purpose, and the instructions emphasize using the CLI's native authentication (OS keychain) rather than manual credential handling.

cloud-provisioning