{"version":"2.1.319","anchor":"context-markdown-deferred-mcp-tools-are-summarised-by-serv","canonical_anchor":"context-markdown-deferred-mcp-tools-are-summarised-by-serv","heading":"Shorter \/context report groups unloaded MCP tools by server","tier":"notice","area":"Context","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319\/e\/context-markdown-deferred-mcp-tools-are-summarised-by-serv","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.319","markdown":"### Shorter `\/context` report groups unloaded MCP tools by server\n\nA shorter form of the `\/context` report lists only loaded MCP tools, totals the rest by server and caps long tables\n\n**Unclear.** Which uses of `\/context` produce the shorter form is not stated.\n\n**What**\n\n`\/context` shows what is using up the conversation's context, the amount of text the model can take in at once. Its text report used to list every MCP tool in one table. MCP tools come from outside programs connected to Claude Code.\n\nThe report now has a shorter form. In that form:\n\n- Only MCP tools that are loaded are listed one by one\n\n- A new \"Other MCP tools, by server\" table shows each server's tool count and \"Tokens if all were loaded\"\n\n- The MCP and agent tables stop after a set number of rows and end with a \"more\" row\n\n**Why**\n\nWith many MCP servers, the report stays short enough to read and still shows what each server would cost if all its tools were loaded.\n\n- Area: Context\n- Names: `\/context`\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}