OpenRouter MCP Server

### Task - Always attach a minimum of 15 relevant scripts or files to each query, providing the LLMs with sufficient input data for thorough analysis. Prioritize comprehensive context to enable ultra-deep thinking and robust outcomes. Do not scan or load full script contents into your own context upfront—reserve this for post-response verification of targeted files only. #### Essential Steps - Instruct each LLM to scan all attached scripts and identify any redundant or duplicated code, as well as any other issues they discover. - Direct each LLM to develop a comprehensive plan to resolve all identified issues, including providing the complete plan in their response. - After receiving a response from an LLM, scan each script identified as requiring changes fully—all at once, in parallel—to verify the accuracy of its claims efficiently. Correct any inaccuracies before proceeding to query the next LLM, incorporating those corrections into the query prompt. - Once the LLMs return their final instructions to you (after completing the query chain), implement the good suggestions to enhance the project and critique the bad ones, ensuring only robust, efficient, and logically sound improvements are applied. #### Additional Instructions - Explicitly remind the LLMs that, while the primary focus is on targeted issues (e.g., redundancy), they should also report and address any other problems they discover. - Always attach a minimum of 15 relevant scripts or files to each query, providing the LLMs with sufficient input data for thorough analysis. Prioritize comprehensive context to enable ultra-deep thinking and robust outcomes. Do not scan or load full script contents into your own context upfront—reserve this for post-response verification of targeted files only.