Use Superagent when an agent workflow needs an explicit safety gate before it accepts user input, calls tools, returns output, or processes a repository. The operator installs the SDK, CLI, or MCP server, configures the API key or local guard model, runs guard checks on incoming messages, redacts sensitive text when needed, scans repositories for agent-targeted risks, and records the resulting classification or scan report as review evidence. Teams can deploy the check in CI, route blocked results into an approval queue, test risky prompts before production use, and review violations before an agent continues. This is not just an SDK listing: the scope boundary is a reviewable safety workflow around agent execution, with clear inputs, decisions, and outputs. Invoke it instead of normal product use when a team needs repeatable guard, redact, scan, or compliance steps that can be embedded in development, CI, or runtime agent operations. Keep each decision traceable to the upstream Superagent result.