Our Commitment to Open Source Chemistry Tools

Why ChemIllusion treats open source as core product philosophy—and how projects like rdkit-agent, our chemistry MCP server, and ChemCP help bring chemistry-aware AI tools to a wider community.

At ChemIllusion, we see open source as part of our product philosophy. Our mission is to help chemists engage with AI, and one way to do that is by making important pieces of our codebase usable beyond the ChemIllusion platform. Increasingly, the most valuable software tools are not closed products, but ecosystems that others can learn from, build on, and adapt.

Some parts of ChemIllusion live directly inside the platform, while others exist as standalone repositories, packages, and agent tools that anyone can use or contribute to. If we build something genuinely useful for chemistry-aware AI, cheminformatics, or chemical education, we want there to be multiple ways for the community to use it beyond our own app.

A good example is rdkit-agent, now available on npm, along with its source repository at scottmreed/rdkit-agent. It is an agent-first cheminformatics CLI powered by RDKit that can validate, convert, and analyze SMILES, SMIRKS, and InChI with structured JSON output. It is designed to work both as a command-line tool and as a Claude Code skill via npx skills add scottmreed/rdkit-agent -g. That makes it useful to a broad audience, from developers to AI-agent builders.

We are also building a chemistry MCP server at chemillusion.com/mcp-server, intended to act as an open infrastructure layer for chemistry-aware agents and applications. It provides a public set of tools for validation, visualization, literature mining, and molecular analysis, alongside authenticated endpoints for higher-cost generative workflows such as image, video, and chemistry generation. The goal is to let other teams add meaningful chemistry capability to their own tools without having to run RDKit infrastructure, external API integrations, or safety guardrails themselves.

One thing that may seem counterintuitive is that the more use ChemIllusion gets, the more open-source work we expect to publish. Open-source projects with a small user base often fade out. Open source tends to thrive when there is an active community using, testing, and improving it. If you have a use case for our MCP or another chemistry-aware integration, we'd love to hear from you.

Our commitment is simple: more parts of ChemIllusion will move into public view over time. More skills, more validators, more chemistry-aware agent tools, more educational experiments, and more reusable infrastructure will be added as the platform grows.

ChemIllusion's web and generator surfaces are also built on RDKit (rdkit/rdkit), Ketcher (epam/ketcher), and Indigo (epam/Indigo) for structure editing and in-browser layout.

If you want to follow that work today, start with rdkit-agent on npm, rdkit-agent on GitHub, and ChemCP (scottmreed/ChemCP). They are different kinds of projects, but together they reflect what ChemIllusion is trying to build: chemistry tools that are not just AI-powered, but inspectable, reusable, and easy to build on.