You must integrate multiple internal APIs and Azure services as tools for several agents, with a standard protocol for tool discovery and invocation rather than one-off SDKs per agent. Which integration approach best fits modern multi-agent design on Azure?
Select an answer to reveal the explanation.
Short Explanation
You want MCP servers and clients to expose tools consistently, with scopes and auth, so agents consume them the same way. Hard-coded per-agent HTTP clients, manual SQL paste, and banning tools don't scale or deliver enterprise value.
Full Explanation
Specifying integration components for agent-to-tool interactions and knowledge routing is an architecture skill; MCP is a listed open standard. Exposing tools through MCP servers/clients with clear tool scopes and authentication standardizes tool ecosystems. Hard-coded unique HTTP clients with divergent auth inside every agent prompt string are brittle. Requiring users to run SQL manually and paste results into chat is manual. Banning all tools and forcing pure generation without enterprise data access destroys capability. Architect tool permission boundaries per agent, observability of tool calls, and versioning of tool schemas.