{
  "id": 205778,
  "title": "5 Underrated MCP Features That Make AI Clients Smarter",
  "url": "https://urgent.news/2026/08/06/5-underrated-mcp-features-that-make-ai-clients-smarter",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-06T05:14:12.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/thatcaffeinateddev/5-underrated-mcp-features-that-make-ai-clients-smarter-1d9k"
  },
  "original_language": "en",
  "account": "Here are the key points from the source:\n\n1. Tool Annotations provide important hints about a tool's behavior, such as readOnlyHint for read-only tools, idempotentHint for tools that don't change environment with each call, destructiveHint for potentially destructive updates, and openWorldHint for tools interacting with external entities. These hints help clients plan execution, present tools to users, and design confirmation flows, though they are not security guarantees.\n\n2. Structured Tool Results allow tools to return data in a JSON format through the structuredContent field. This structured data can be more easily validated, rendered, and passed around compared to plain text results. It's important to also return the JSON data in a text content block for backward compatibility.\n\n3. Output Schemas define the expected structure of structuredContent before and after a tool invocation. This allows clients to validate results, use typed integrations, and present clearer tool documentation. The inputSchema defines what the client must provide, and the outputSchema defines what the client can expect. Clients are encouraged to validate the results against the output schema.\n\n4. Progress Reporting is available for long-running requests. Clients can include a progressToken in the request metadata, and the server can send progress notifications back to the client with a percentage completed and a message. This provides users with updates on the progress of time-consuming operations like indexing repositories or running large AI workflows.",
  "summary": "Everyone's building MCP servers. Most conversations focus on transports, authentication, OAuth, and exposing tools. Those topics are important, but while building a few MCP servers recently, I found myself appreciating a different part of the specification that doesn't seem to get nearly as much attention. The interesting thing is that these aren't hidden APIs. They're already part of the Model…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}