deepmirror documentation
deepmirror helps you train predictive models on your own assay data, score new molecules, and design the next round of compounds. Use it from the web app, from Claude or OpenAI Codex over MCP, or directly through the REST API.
Guides
Models How deepmirror trains and scores predictive models, and how to read the numbers reported in the app. How models are evaluated Which molecules your model is scored on, how they are chosen, and what the numbers mean.Integrations
MCP Use deepmirror from Claude, OpenAI Codex, or your own application via the remote MCP server. Connecting to MCP manually Register the deepmirror MCP server in Claude, Claude Code, OpenAI Codex, or the OpenAI Responses API. Example MCP workflow A single prompt that runs upload, review, training, goals and generation end to end.Reference
API Reference
Browse the REST API for every endpoint, request schema and response.
Changelog
See what's new in deepmirror, including product updates and changes to how results are reported.