# Anthropic Just Fixed MCP’s Biggest Problem Source: https://www.youtube.com/watch?v=TVOoMwkpSRQ Recap page: https://rapidrecap.app/video/TVOoMwkpSRQ Generated: 2026-01-15T14:35:18.237+00:00 --- ## Quick Overview Anthropic fixed a major context consumption issue in the Model-Client-Protocol (MCP) system by introducing 'Tool Search,' which dynamically loads only necessary tool definitions on demand, reducing context usage by up to 85% compared to preloading all tools. **Key Points:** - The primary problem solved was excessive context window usage caused by preloading too many MCP tool definitions, exemplified by the GitHub MCP consuming 46k tokens across 91 tools. - The solution, 'MCP Tool Search,' allows Claude Code to dynamically load tools into context only when needed, resulting in an 85% token reduction for tool definitions. - Tool Search operates via a 10% context threshold trigger, dynamically loading 3-5 relevant tools at once, rather than preloading all definitions. - Server developers must add the 'advanced-tool-use-2025-11-20' beta header to enable this feature. - Server instructions should guide Claude on tool execution order (e.g., check PR status before reviewing PR) to maintain correct workflows. - Tool descriptions require optimization (short, function-leading, keyword-rich, constraints in input schema) to maximize search effectiveness. - The implementation checklist involves 8 steps, including adding the search tool (Regex or BM25 variant), marking non-essential tools with 'defer_loading: true', and monitoring context usage. ![Screenshot at 00:04: A diagram illustrating the context usage breakdown in Claude Sonnet, showing that 'MCP tools' consume 44.6% \(89.1k tokens\) of the 200k token context window before any user message, highlighting the severity of the preloading problem.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-04.jpg) **Context:** The video addresses a significant scaling challenge within the Model-Client-Protocol (MCP) framework used for connecting large language models (like Claude) to external tools. As MCP servers accumulate more tools (up to 50+), the tool definitions consume a large portion of the model's limited context window, making complex tasks inefficient or impossible before any actual conversation begins. Anthropic introduced 'Tool Search' to mitigate this context pollution. ## Detailed Analysis The video details the introduction of 'MCP Tool Search' designed to solve the massive context consumption problem caused by preloading extensive tool definitions in the MCP framework. Initially, an example context usage screen shows that MCP tools can consume 89.1k tokens (44.6% of a 200k window) before any user interaction, which is problematic, especially for services like GitHub with 91 available tools. Tool Search addresses this by loading tool definitions dynamically (on-demand) instead of preloading everything. This dynamic loading is triggered when tool descriptions exceed a 10% context threshold, loading only 3-5 relevant tools at once, leading to an 85% token reduction in context usage for tool definitions. Implementation requires client developers to enable the beta header 'anthropic-beta: "advanced-tool-use-2025-11-20"' in API requests. Server developers must add the new search tool (either Regex-based for structured naming or BM25 for semantic search) to their tool list and mark non-essential tools with '"defer_loading": true' to prevent immediate loading. Additionally, server instructions should be used to guide Claude on tool execution order (e.g., checking PR status before reviewing a PR). Best practices for tool descriptions include leading with function, keeping them 1-2 sentences, adding searchable keywords, and placing constraints in the 'input_schema' rather than the description, as every word costs tokens. The overall implementation involves an 8-step checklist culminating in monitoring context usage. ### MCP Context Problem - GitHub's MCP server consumes a staggering 46k tokens across 91 tools that cannot be individually enabled or disabled - This consumes nearly a quarter of Claude Sonnet/Opus 4's context window before any code is written - For comparison, an optimized toolset uses only 1.6k tokens. ### The Solution - MCP Tool Search - Allows Claude Code to dynamically load tools into context when needed, instead of preloading all definitions - Achieves 85% token reduction, only loading 3-5 tools at once when the 10% context threshold is triggered. ### Server Implementation Steps - 1. Enable Beta Header ('anthropic-beta': 'advanced-tool-use-2025-11-20') - 2. Add Tool Search Tool (Regex or BM25) - 3. Mark non-essential tools with '"defer_loading": true' - 4. Keep 3-5 essentials without deferred loading (set 'defer_loading': false) - 5. Optimize tool descriptions - 6. Add server instructions to guide workflow order - 7. Test with 10+ tools - 8. Monitor context usage. ### Tool Search Variants - Regex tool search is best for structured naming (e.g., 'get_current_date') - BM25 tool search is best for semantic search (e.g., 'tools for weather'). ### Tool Description Best Practices - Lead with function (what it does, first) - Keep to 1-2 sentences - Add searchable keywords & synonyms - Put constraints in input_schema, not the description - Every word costs tokens. ![Screenshot at 00:00: Diagram showing the MCP standardized protocol connecting AI applications to data sources and tools, illustrating the bidirectional data flow.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-00.jpg) ![Screenshot at 00:04: Context Usage visualization showing 44.6% of the 200k context window consumed by MCP tools before any user message.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-04.jpg) ![Screenshot at 00:09: A list of available MCP tools, specifically showing numerous GitHub-related functions, illustrating the sheer volume of definitions being preloaded.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-09.jpg) ![Screenshot at 00:25: Appendix section detailing that the GitHub MCP consumes 46k tokens across 91 tools, contrasting it with an optimized toolset using only 1.6k tokens.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-25.jpg) ![Screenshot at 00:57: Title slide for the solution, 'MCP Tool Search,' which promises to cut context usage by 85%.](https://ss.rapidrecap.app/screens/TVOoMwkpSRQ/00-00-57.jpg)