Recently, a guide named "Writing Effective Tools for LLM Agents—Leveraging LLM Agents" has been made available. Its primary objective is to aid developers in creating efficient tools for LLM Agents through the utilization of MCP. The guide introduces a three-phase iterative methodology: 'prototyping-evaluation-collaboration'. Moreover, it delineates five fundamental design tenets. These include the judicious selection of tools, the preservation of distinct namespaces, the refinement of tool context return values, the prioritization of Token efficiency, and the elevation of tool description quality via prompt engineering techniques. Notably, a significant portion of the insights presented in the guide was autonomously generated by Claude Code and subsequently validated against a reserved test set to mitigate the risk of overfitting. Furthermore, the guide offers an open-source tool evaluation Cookbook and states that as the MCP protocol and the foundational LLMs continue to evolve, the capabilities of these tools will experience further enhancements.