The Agent Skills Directory

Command Execution: The skill utilizes the Bash tool to execute a local CLI binary (ms). This tool is used to authenticate the user, list available connectors, and retrieve metadata about environment policies. The usage is restricted to informational commands such as list, show, and list-actions.
Workspace Data Processing: The skill reads local workspace files, specifically ms.config.json and memory-bank.md, to determine the current project state and environment context. This is a standard operation for identifying which data sources are already integrated into a project.
Indirect Prompt Injection Surface: Because the skill ingests data from external sources (the CLI tool output and ms.config.json), there is a theoretical surface for indirect prompt injection if these sources were compromised with adversarial content. The risk is minimized as the skill primarily processes structured JSON data and uses jq for filtering, which helps maintain data integrity before presentation to the agent.

list-connectors