Reexpress MCP Server is a drop-in solution to add state-of-the-art statistical verification to your complex LLM pipelines, as well as everyday use of LLMs for search and QA in software development and data science settings. It's the first reliable, statistically robust AI second opinion for your AI workflows. Simply install the MCP server and append the Reexpress prompt to the end of your chat text. The tool-calling LLM (e.g., Claude Opus 4.1) will then check its response with the provided pre-trained Reexpress SDM estimator, which ensembles gpt-5-2025-08-07, gemini-2.5-pro, and granite-3.3-8b-instruct (run locally), along with the LLM output, and computes a robust estimate of predictive uncertainty against a training/calibration database from OpenVerification1. You can adapt the model to your tasks by calling ReexpressAddTrue or ReexpressAddFalse after a verification, and future calls will incorporate those updates when calculating the verification probability. The package includes training scripts for retraining or using alternative underlying LLMs. Data is processed locally; only standard LLM API calls to Azure/OpenAI and Google are made, with optional web-search via your MCP client. File access is controlled via ReexpressDirectorySet() and ReexpressFileSet(). Version 2.0.0 updates the SDM estimator and code for efficiency, with canonical SDM as the reference.