{"version":"2.1.301","anchor":"mcp-call-can-skip-model-processing-with-result-not-for-model","canonical_anchor":"mcp-call-can-skip-model-processing-with-result-not-for-model","heading":"mcp_call describes a way to keep tool results out of Claude's processing","tier":"notice","area":"SDK","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301\/e\/mcp-call-can-skip-model-processing-with-result-not-for-model","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.301","markdown":"### mcp_call describes a way to keep tool results out of Claude's processing\n\nThe description of the mcp_call request now says results skip the usual model-side processing when result_not_for_model is set\n\n**Unclear.** Only the request's description mentions result_not_for_model; whether requests already accept it is not confirmed.\n\n**What**\n\n`mcp_call` is a request a program controlling Claude Code can send to call a tool on an MCP server (an add-on that provides tools) directly. Its description used to say the result always goes through the same processing as when Claude itself calls the tool. It now says this happens unless `result_not_for_model` is set.\n\n**Why**\n\nA program making these calls has an option to get a result that is not prepared for Claude to read.\n\n- Area: SDK\n- Names: `mcp_call`, `result_not_for_model`\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 2\/5\n- Scope: individual\n- Heads-up: no"}