Book a demo

How to Set Up PatSnap Chemical Molecular MCP

This tutorial installs PatSnap Chemical Molecular MCP in Claude Desktop, Cursor, or Continue.dev. After setup, ask your AI to search patents by chemical structure, predict ADMET properties, or extract every compound from a patent — in plain language, without leaving your AI conversation.

What You Will Accomplish

You’ll add a connection between your AI environment and PatSnap’s chemical intelligence database. MCP (Model Context Protocol) is an open standard that lets AI assistants query external databases live during a conversation. After setup, you can ask your AI to search compound structures, predict ADMET properties, or extract every molecule from a patent—all in plain language.

Who Built This MCP

This guide uses PatSnap Chemical Molecular MCP—built by PatSnap, the patent intelligence platform indexing 208M+ patents across 174 jurisdictions and 282M+ compounds. The same database powers enterprise chemical search and freedom-to-operate workflows used by pharmaceutical and biotech R&D teams worldwide. Updates pull from USPTO, EPO, CNIPA, and WIPO daily.

Why Use This MCP

Once installed, your AI can handle structure-based queries without leaving the conversation. Ask it to find compounds similar to a SMILES string—it returns analogs with Tanimoto scores and patent status, helping you identify white space or infringement risks before synthesis. Request ADMET predictions for a lead compound—it outputs bioavailability, clearance, and toxicity flags, informing go/no-go decisions in early screening. Need the Maximum Common Substructure across a series? Paste your SMILES list, and the MCP calculates the shared scaffold for SAR analysis. Extract all compounds from a patent filing by number, or batch-fetch up to 100 structures by InChIKey for library annotation.

Prerequisites

Before you start, confirm you have:
  • MCP-compatible AI environment: Claude Desktop, Claude Code Editor, or Cursor. Download from claude.ai or cursor.com.
  • PatSnap account: Sign up at open.patsnap.com—free tier includes 10,000 credits, no credit card.
  • API key: Generate from the MCP marketplace after login.

Step 1: Get Your API Key

Log in to PatSnap Open Platform. Navigate to the MCP marketplace and locate Chemical Molecular MCP. Click Generate API Key—the platform displays a long alphanumeric string. Copy it. You’ll paste this into your configuration file in the next step.Don’t close the browser tab yet. You’ll return here later to monitor credit usage if you want to track consumption.

Step 2: Add the MCP to Your Configuration File

Your AI client needs to know where to find the MCP. Open your MCP configuration file: Claude Desktop (macOS) ~/Library/Application Support/Claude/claude_desktop_config.json, Claude Desktop (Windows) %APPDATA%\Claude\claude_desktop_config.json, Cursor .cursor/config.json in your project root, Claude Code Editor .claude/config.json in your workspace.If the file doesn’t exist, create it. Paste this JSON, replacing YOUR_API_KEY with the key you copied:
{
  "mcpServers": {
    "chemical_molecular": {
      "url": "https://connect.patsnap.com/713886/logic-mcp?apikey=YOUR_API_KEY",
      "type": "streamableHttp"
    }
  }
}
Save the file. If you already have other MCPs configured, add the "chemical_molecular" block inside the existing "mcpServers" object—don’t replace what’s there.Restart your AI client. Claude Desktop and Cursor require a full quit-and-relaunch for MCP changes to load.

Step 3: Verify the Connection

Give it a try: “Find compounds similar to omeprazole (SMILES: COc1ccc2nc(S(=O)Cc3ncc(C)c(OC)c3C)[nH]c2c1) with Tanimoto above 0.7.” If you see results listing analogs with similarity scores and patent numbers, you’re connected.
What you’ll see: The MCP returns a table of structurally similar proton pump inhibitors, each with a Tanimoto score, assignee, and legal status. Most major PPI patents have expired—except for newer salt or formulation variants like esomeprazole magnesium in US7678902B2. This confirms structure search, patent integration, and legal-status filtering are all working.

Example Prompts

Try these queries to explore what the MCP can do:“Search for aspirin (CC(=O)OC1=CC=CC=C1C(=O)O) and show me compounds with Tanimoto similarity above 0.8.”Returns salicylate analogs from PubChem and patent databases with similarity scores, patent coverage, and assignees.“Predict ADMET properties for ibuprofen: CC(C)CC1=CC=C(C=C1)C(C)C(=O)O.”Outputs oral bioavailability, CYP metabolism flags, hERG risk, and blood-brain barrier permeability scores.“What is the maximum common substructure for these three SMILES: [your SMILES list]?”Calculates the largest shared scaffold across your compound set—useful for identifying the core structure in a SAR series.“Extract all compounds from patent US7576219B2.”Lists every chemical structure disclosed in the filing, with SMILES and InChIKeys for downstream analysis.

Troubleshooting

SymptomCauseFix
AI doesn’t respond to chemical queriesMCP not loaded after config changeFully quit and restart your AI client—minimize doesn’t reload MCPs
“API key invalid” errorKey copied incorrectly or expiredRegenerate key in MCP marketplace and update config file
Results missing patent numbersQuery returned pre-patent compoundsRefine similarity threshold or try a different scaffold—some analogs predate modern patent filings
ADMET predictions return “N/A”Model cannot predict for this structure classADMET-AI model has coverage gaps for peptides, organometallics, and some natural products

Next Steps

Explore other MCPs in the marketplace: Add Biology Modality MCP for biosequence search or Pharma Intelligence MCP for clinical trial and pipeline data.Set up batch workflows: Use the InChIKey batch fetch tool to pull structure data for up to 100 compounds at once from screening libraries.Combine with ChemDraw: Export SMILES from ChemDraw, paste into your AI conversation, and get instant patent landscape or ADMET reports without switching tools.

Ready to Connect the MCP?

Start free—10,000 credits, no credit card, no subscription.→ Get Your API Key—sign up at open.patsnap.com→ Add the MCP to Your AI—find it in the marketplace→ Try in Browser—PatSnap Eureka, no install required
Note: Information based on publicly available sources as of 2026. Product features may change. Contact PatSnap for current capabilities and pricing.

Eureka built for innovation research

Eureka built for research
Domain-specific AI agents for IP, Engineering, Life Sciences, and Materials
Patents, Scientific Literature, Compounds & More Unified in One Platform
Ask, Research, Solve, Draft, and Validate Your Work from Weeks to Minutes
Try it for Free

Your Agentic AI Partner
for Smarter Innovation

PatSnap fuses the world’s largest proprietary innovation dataset with cutting-edge AI to
supercharge R&D, IP strategy, materials science, and drug discovery.

Book a demo