{"version":"2.1.301","anchor":"mcp-call-results-can-be-marked-not-for-model-with-a-too-lar","canonical_anchor":"mcp-call-results-can-be-marked-not-for-model-with-a-too-lar","heading":"MCP tool results can be kept from the model, with a size limit","tier":"use","area":"SDK","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301\/e\/mcp-call-results-can-be-marked-not-for-model-with-a-too-lar","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301","markdown":"### MCP tool results can be kept from the model, with a size limit\n\nAn MCP tool call made through the control channel can mark its result as not for the model, and too-large results now return an error text\n\n**Unclear.** It is not clear what turns the not-for-model setting on or whether any program sends it yet.\n\n**What**\n\nMCP servers are external programs that give Claude Code extra tools. When a program that embeds Claude Code asks it to call an MCP tool directly, the request can now carry a setting saying the result is not meant for the model. If the result is too large, Claude Code sends back a fixed error message instead of the result itself.\n\nBefore, results were always passed back in full, with no size check.\n\n**Why**\n\nPrograms that embed Claude Code through the SDK (the toolkit for building on Claude Code) or run it remotely may now get an error message back instead of the content of a large MCP tool result.\n\n- Area: SDK\n- Tier: Use it now\n- Useful: 3\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}