CyberRXN documentation
Find reaction evidence, check chemistry and build a traceable workflow.
CyberRXN combines a reaction corpus with chemical checks and tools for assistants. Use it to find a reported preparation, inspect its original evidence, validate an equation, or connect a chemistry workflow to an application.
Choose your starting point
| Goal | Start here |
|---|---|
| Use ChatGPT or Claude | Connect an assistant |
| Make your first request | Quickstart |
| Work in Python | Python SDK |
| Work in JavaScript or TypeScript | JavaScript SDK |
| Work in a terminal | CLI reference |
| Check or balance an equation | Chemistry validation |
| Convert molecular representations | Conversion |
| Control API-key usage | Limits and reset periods |
The evidence boundary
A retrieved observation is a source record. A structure match, a model prediction, a solved balance and a reviewed source are separate facts. Source conditions can be incomplete or extracted from the wrong stage. Inspect the original paragraph before treating a number as a reaction condition.
Chemical validation checks consistency, not experimental feasibility. Unknown pressure is not atmospheric pressure; a model-added molecule is not a reported reagent.
Public service endpoints
- Dashboard: app.cyberrxn.com
- REST API:
https://api.cyberrxn.com/v1 - MCP:
https://mcp.cyberrxn.com/mcp - Current guide: reaction-evidence Markdown
- Schema: OpenAPI
For automated readers, use llms.txt, llms-full.txt, or each page's Copy Markdown action. Documentation is versioned with its contract snapshot; release notes explain availability.