The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by ingesting untrusted data from network devices. \n- Ingestion points: The get_node_console_log and execute_command tools retrieve data directly from external nodes. \n- Boundary markers: No explicit delimiters or instructions to ignore embedded content are present in the skill instructions. \n- Capability inventory: The skill has extensive control over the environment, including tools to start/stop nodes, wipe configurations, and set new startup configurations. \n- Sanitization: No evidence of sanitization or validation of the retrieved node output was found. \n- [COMMAND_EXECUTION]: The execute_command tool grants the agent the ability to run arbitrary CLI commands on remote network devices. While intended for operations, this is a significant privilege. \n- [EXTERNAL_DOWNLOADS]: The skill specifies the use of the cml-mcp Python package and the pyats extra, which are external dependencies required for execution.

cml-node-operations