browsing-capabilities/server-detail-view

Server Detail View & Token Estimator

Connection status, the full capability list, and a context-budget card estimating the token cost of a server's tool/resource/prompt definitions.

Selecting a server in the sidebar opens its detail view — a single place to see everything about that server’s connection and capabilities at a glance.

What’s in the detail view

  • Connection status and actions — current state (connected/disconnected/connecting), plus controls to disconnect, reconnect, or remove the server.
  • Full capability list — every tool, resource, and prompt the server exposes, without needing to expand the sidebar tree.
  • Auth panel (OAuth-protected servers only) — token status, expiry, and scopes; see Authentication & OAuth 2.1.
  • Context-budget card — see below.

Context-budget card

The context-budget card estimates how many tokens the server’s tool, resource, and prompt definitions would consume if loaded into a model’s context window — before you’ve even made a single call.

This matters because every tool/resource/prompt a client connects to has a schema and description that gets sent to the model as part of its context, whether or not the model ever calls it. A server with many verbose tool definitions can quietly eat a large chunk of a model’s context budget.

The card breaks this down segment by segment:

  • Cost per tool definition (name, description, input schema)
  • Cost per resource definition
  • Cost per prompt definition
  • A running total for the whole server

Why it’s useful

  • Designing servers — if you’re authoring an MCP server, this tells you how “expensive” your tool definitions are to a model before you ship them, so you can trim verbose descriptions or overly complex schemas.
  • Evaluating third-party servers — before wiring an external server into a production agent, check how much of your context budget it will consume just by being connected, independent of actual usage.
  • Comparing servers — if choosing between two servers with similar functionality, the context-budget card gives a concrete number to compare.
  • This estimates the cost of definitions only. The cost of an individual tool call’s response is shown separately — see Per-Call Token Footprint.
  • Estimation uses gpt-tokenizer; actual token counts may vary slightly by the model you ultimately use in production.

Edit this page on GitHub