Chemical space,
one prompt away.
Ask ChatGPT, Claude or your own agent for purchasable analogs. DeepMedChem makes the whole purchasable chemical space searchable, built for 100T+ molecules, by shape, electrostatics and fingerprints, right inside the chat.
https://mcp.deepmedchem.com/mcpNo account · No API key · Claude Code, Codex and any MCP client
ChatGPT · published plugin
Claude Code · two commands
×DeepMedChem
Similarity search across trillions of make-on-demand molecules by Morgan, shape or electrostatics, with the skill that keeps Claude honest about the hits.
From prompt to flask
The whole purchasable chemical space
We are making every purchasable molecule searchable. The platform is built for 100T+ molecules, so an agent can find what it needs, price it and have it made for cheap in a few weeks.
Search trillions
Your agent queries the major make-on-demand spaces by shape, electrostatics or fingerprints, in seconds.
Price and pick
Hits come back with list prices and vendor catalogue IDs. Most spaces report synthesis success above 80%.
Order
Make-on-demand molecules arrive synthesized in 2–8 weeks; in-stock catalogues ship right away.
- Synple eXplore9.5T
- Synple7.6T
- ChemInfinita794B
- Freedom Space296B
- Enamine REAL93B
- VAST6.8B
- D2B SpaceM11.5B
- Onepot CORE
- + in-stock catalogues
Sizes, lead times and success rates per space: database breakdown. Ordering and quotes via dmc order.
Coding agents & CLI
Terminal-native, too
Add DeepMedChem to Claude Code or Codex in one command, then ask for analogs between commits.
Claude Code
# inside a Claude Code session: plugin + skill
> /plugin marketplace add Deep-MedChem/plugin> /plugin install deepmedchem@deepmedchem
# or only the MCP server, from your shell
$ claude mcp add --transport http deepmedchem https://mcp.deepmedchem.com/mcp
Codex CLI
# add the MCP server
$ codex mcp add deepmedchem --url https://mcp.deepmedchem.com/mcp
# optional: add the skill
$ mkdir -p ~/.codex/skills/deepmedchem$ curl -fsSL https://www.deepmedchem.com/skill.md -o ~/.codex/skills/deepmedchem/SKILL.md
Then just ask: “find purchasable analogs of CC(=O)Oc1ccccc1C(=O)O in Enamine REAL”
Gemini CLI: gemini mcp add -t http deepmedchem https://mcp.deepmedchem.com/mcp
Anywhere MCP runs
Bring it to your stack
Chat apps, IDEs and your own code: one hosted server, the same three tools.
Settings → Connectors → Add custom connector, then enable it under + → Connectors in a new chat. Works on web and desktop.
- Name
- DeepMedChem
- URL
- https://mcp.deepmedchem.com/mcp
- Authentication
- No sign in
Agent-readable
Built for agents, not just chats
Point any agent at our skill and it will connect, search and cite results on its own.
Read https://www.deepmedchem.com/skill.md and follow it to search chemical space with DeepMedChem.
llms.txt
A plain-text map of Deep MedChem for language models: products, docs and how to connect.
/llms.txt
skill.md
A drop-in agent skill: connection recipes (MCP or plain HTTP) and the rules for good searches.
/skill.md
MCP server
Streamable HTTP, stateless JSON-RPC, no auth. Tool schemas are live via tools/list.
mcp.deepmedchem.com/mcp
Under the hood
Three tools. Zero setup.
catalog
Databases, sizes, prices and guest limits.
similarity_search
One SMILES, one space, up to 20 hits by Morgan, shape or ESP.
render_results
Re-open earlier hits as an interactive grid, without searching again.
Hosts with MCP Apps show an interactive structure grid with prices, vendor IDs, CSV export and a link to the saved CHEESE job. Similarity scores describe structure, not biological activity or safety. Need substructure search, filters or bigger result sets? Use CHEESE or the Python SDK.