A fire department agent must consult a radio-log knowledge store and call a CAD status API. Which Foundry design best combines those needs?

A fire department agent that must consult a radio-log knowledge store and call a CAD status API needs both capabilities in one Foundry agent design: knowledge retrieval for radio-log history and an API tool for live CAD status. Historical logs provide narrative and prior-context grounding; CAD provides current operational state. Combining knowledge and tools is the standard pattern when municipal incident support needs yesterday’s record and today’s status together.

Knowledge store for radio-log history plus an API tool for live CAD status in one agent batch index run fails because a CAD status tool must be invoked live at runtime, not during a batch index run that builds knowledge; indexing cannot provide current operational status. Knowledge store for radio-log history plus an API tool for archived CAD status in one agent design fails because CAD status must be current to support operations, so an archived snapshot cannot substitute for a live API call. Knowledge store for radio-log history plus an API tool for live CAD status in one agent prompt template fails because a prompt template alone does not provide knowledge retrieval and tool invocation; both capabilities need to be wired into one agent design, not just written as template text.

Exam caveat: tool auth to CAD and retention/access controls on radio-log knowledge are mandatory design items, not optional extras. Operational check: in a drill, verify the agent retrieves a known log passage and successfully calls a CAD status tool (or mock), and confirm degraded behavior is explicit when either source is unavailable.