Quick Start

NovoMCP connects to any MCP-compatible AI client — Claude, ChatGPT, Cursor, Windsurf, GitHub Copilot, and more. One URL, one authentication step. Your AI gains molecular intelligence without new software.

Prerequisites: A NovoMCP API key (get started free at app.novomcp.com) and an MCP-compatible AI client.

Step 1: Connect Novo (all users)

Novo gives you discovery, profiling, ADMET, compliance, and research tools.

Claude Desktop / Web / Mobile:
1. Go to Settings > Connectors
2. Click "Add custom connector"
3. Enter URL: https://ai.novomcp.com/mcp
4. Click Connect — a login page opens in your browser
5. Enter your NovoMCP API key (nmcp_...)
6. Click Authorize Access — done!

Claude Code (Terminal):
1. claude mcp add --transport http novo https://ai.novomcp.com/mcp
2. claude
3. /mcp  → authenticate in browser

Step 2: Connect Novo Compute (paid plans)

Novo Compute adds quantum chemistry, molecular dynamics, docking, conformer search, neural network potentials, and structure prediction. Available on Core, Team, and Enterprise plans.

Claude Desktop / Web / Mobile:
1. Go to Settings > Connectors
2. Click "Add custom connector"
3. Enter URL: https://compute.novomcp.com/mcp
4. Click Connect — enter your Novo Compute API key (ncmcp_...)

Claude Code (Terminal):
1. claude mcp add --transport http novo-compute https://compute.novomcp.com/mcp
2. claude
3. /mcp  → authenticate in browser

Generate your Novo Compute API key at app.novomcp.com/keys. Requires a paid plan — upgrade at app.novomcp.com/billing.

Team Setup (.mcp.json)

For teams, add both servers to your repo's .mcp.json so every team member gets them automatically:

{
  "mcpServers": {
    "novo": {
      "type": "http",
      "url": "https://ai.novomcp.com/mcp"
    },
    "novo-compute": {
      "type": "http",
      "url": "https://compute.novomcp.com/mcp"
    }
  }
}

Each team member authenticates individually on first use. Auth tokens refresh automatically.

Cursor / Windsurf / Other MCP Clients

Novo:         https://ai.novomcp.com/mcp
Novo Compute: https://compute.novomcp.com/mcp
Transport:    Streamable HTTP
Auth:         OAuth 2.0 (automatic) or API key header

Try it

Once connected, ask your AI:

"What's the ADMET profile of aspirin? Is it hepatotoxic?"

Authentication

Two ways to authenticate — both work with any client.

OAuth 2.0 (Recommended)

When you add NovoMCP as a connector, your client handles OAuth automatically. You'll be redirected to a login page, enter your API key, and you're done. Tokens refresh on their own.

API Key Header

For programmatic access or custom integrations, pass your API key in the X-API-Key or Authorization: Bearer header.

Key Format: Novo keys: nmcp_[random]. Novo Compute keys: ncmcp_[random]. Both draw from the same credit balance. Store securely — keys cannot be retrieved after creation.

Tool Reference

43 tools, 5 resources, and 5 prompts. Your AI selects and chains them automatically — you describe what you need in plain language.

Molecular Intelligence

get_molecule_profileFull molecular dossier — properties, ADMET predictions, and regulatory compliance. Pre-computed for 122M+ compounds, retrieved in under 50ms.

get_molecule_infoBasic molecular properties via RDKit: molecular weight, LogP, TPSA, hydrogen bond donors/acceptors, rotatable bonds.

calculate_propertiesOn-demand RDKit property calculation: MW, exact mass, LogP, LogD, TPSA, SA Score, Lipinski/Veber violations, QED.

get_3d_properties32+ 3D molecular properties from conformer generation: geometry, energy, electrostatics, surface/volume.

search_similarFind structurally similar molecules using Tanimoto similarity. Returns top matches with similarity scores.

filter_moleculesFilter database by property ranges. Supports molecular weight, LogP, QED, TPSA. Exclude controlled substances.

batch_profileProfile multiple molecules in a single request. Up to 100 compounds per call.

optimize_moleculeGenerate optimized variants via NVIDIA MolMIM targeting specific QED, LogP, and similarity profiles. Every output compliance-screened before delivery.

ADMET & FAVES Compliance

FAVES screens every molecule against eight regulatory jurisdictions — DEA, FDA, EPA, CWC, EU REACH, BTWC, Australia, OPCW — inline with every prediction, not as a separate step.

predict_admet31 ML models covering absorption, distribution, metabolism, excretion, and toxicity — including CYP450 substrate/inhibitor status, cardiotoxicity, and hepatotoxicity.

check_complianceFull FAVES regulatory assessment with context-aware evaluation for intended use case.

screen_libraryScreen compound libraries up to 1000 molecules with optional full FAVES assessment and filtering.

Structure Prediction

get_protein_structureSmart protein structure resolver. Accepts names (EGFR, CDK2), PDB IDs (1M17), or sequences. Returns experimental structures or OpenFold3 predictions.

predict_structurePredict 3D protein structure from amino acid sequence via OpenFold3. Async — returns job ID.

get_structure_resultRetrieve completed structure predictions. Check status and download PDB files with confidence scores.

Literature & Data Search

search_literatureSemantic search across 14,000+ curated drug discovery papers. Returns titles, authors, DOIs, relevance scores.

search_patentsSearch 2,400+ USPTO pharmaceutical patents. Returns patent numbers, applicants, abstracts.

search_biorxivSearch bioRxiv/medRxiv preprint servers for latest research. 250k+ preprints with full metadata.

search_chemblSearch ChEMBL database for compounds, targets, and bioactivity. 2.4M compounds with SMILES and activity data.

search_clinical_trialsSearch ClinicalTrials.gov for clinical studies. 500k+ trials with status, phase, and enrollment.

Omics & Target Discovery

target_discoveryIdentify drug targets from 108,000 omics-derived target-disease associations. Returns dockable targets ranked by evidence score with suggested PDB structures.

stratify_patientsPharmacogenomic patient stratification across 56 pharmacogene profiles. Predicts metabolizer phenotypes, dose adjustments, and population-level viability.

Property Prediction

predict_pkaPredict acid dissociation constant (pKa) for ionizable groups. Custom Chemprop D-MPNN trained on 22,726 IUPAC measurements. Critical for charge state at physiological pH.

predict_solubilityPredict aqueous solubility (LogS) with temperature dependence (273-368K). Custom Chemprop model trained on AqSolDB + MixtureSolDB. Returns mg/mL and solubility category.

predict_bdePredict bond dissociation energies (kcal/mol) for all C-H bonds. Identifies metabolic soft spots — lower BDE means higher susceptibility to CYP-mediated oxidation.

Quantum Chemistry & Neural Potentials

run_qm_calculationGFN2-xTB semi-empirical QM: single-point energy, geometry optimization, solvation free energy (ALPB). Returns HOMO-LUMO gap, dipole, optimized geometry.

run_conformer_searchGenerate conformer ensemble via CREST (GFN2-xTB) or RDKit ETKDG. Returns ranked conformers with Boltzmann populations. Essential before docking.

dock_with_strainCalculate strain energy of a docked ligand pose using xTB. High strain (>5 kcal/mol) indicates the docking score may be an artifact. Use after dock_molecules.

compute_energyFast energy computation via ANI-2x (organic molecules) or MACE-MP-0 (universal). ~100x faster than xTB. Ideal for batch conformer ranking and screening.

GPU Compute Pipeline

lead_optimizationScaffold hopping via RDKit substructure replacement. Generates structurally novel variants enriched with ADMET and compliance scoring.

dock_moleculesGPU-accelerated molecular docking via AutoDock-GPU. Dock compounds against protein targets, get binding affinities and pose analysis.

run_molecular_dynamicsGPU molecular dynamics via GROMACS. Simulate protein-ligand binding stability over nanosecond timescales.

Pipeline & Context

save_funnel_contextPersist discovery pipeline state for session resumption across long-running campaigns.

get_funnel_contextResume a discovery pipeline from any saved checkpoint — continue from any step without re-running prior work.

save_funnel_stageLog a discovery funnel stage with full audit trail for GxP reproducibility.

get_funnel_auditRetrieve complete funnel audit log — every decision, tool call, and human input for peer review.

get_pipeline_auditPer-molecule audit trail for pipeline executions: SMILES validation, tool results, dispositions, exclusion reasons.

list_jobsList async pipeline jobs with status, progress, and timestamps. Filter by service or status.

get_job_statusCheck status of async jobs (MD simulation, docking, structure prediction) with result details.

generate_dynamicsGenerate AI-accelerated conformational ensemble from a protein structure.

Data Connectors & Platform

pull_from_sourceBidirectional data pipeline: pull compounds from Databricks/Snowflake, run ADMET + FAVES processing, push enriched results back.

push_to_destinationList connections, discover schemas, preview mappings, export results to Snowflake, BigQuery, Databricks, Supabase.

get_platform_infoPlatform information: database statistics, ADMET capabilities, compliance lists.

get_credit_usageCheck your credit balance, usage statistics, and research value realized.

Resources

5 resources providing reference data — compliance schedules, ADMET endpoints, database statistics, and platform changelog.

compliance_schedulesDEA Schedule I-V, CWC chemical weapons lists, FDA banned substances, EPA PBT chemicals, EU REACH restricted compounds.

admet_propertiesList of 40+ ADMET property predictions available: absorption, distribution, metabolism, excretion, toxicity endpoints.

tier_featuresFeatures available at each subscription tier with tool access details.

database_statsCurrent database statistics: 122M indexed molecules, 14,398 curated papers, 2,416 USPTO patents.

changelogRecent changes to NovoMCP tools, features, and API.

Prompts

5 pre-built prompts for common workflows. Your AI can also chain tools freely in response to natural language.

quick_checkFast safety screening — controlled substance status, structural alerts, and compliance flags.

full_analysisComplete molecular workup — properties, ADMET, compliance, drug-likeness, and recommendations.

find_alternativesFind structurally similar molecules that pass compliance — useful when a lead is flagged.

literature_reviewSearch papers, patents, and clinical trials for a target or topic.

discovery_funnelThe full autonomous pipeline — target identification, lead optimization, GPU docking, MD simulation, and patient stratification. One prompt.

Enterprise Connectors

Pull compounds from your warehouse, enrich with molecular intelligence, push results back. Four connector adapters with OAuth sign-in, schema discovery, AI-assisted field mapping, and full audit logging.

Snowflake

Databricks

BigQuery (OAuth)

Supabase

How It Works

1. Discover

discover_schema introspects the target system — tables, columns, types — and normalizes to 5 standard types.

2. Map

preview_mapping aligns NovoMCP tool output to target columns. Template, auto-matching, or AI-assisted mapping.

3. Export

export_results writes data through the appropriate connector adapter.

Governance: All connector credentials are stored in Azure Key Vault — never in the database. All export operations are audit-logged with full traceability.

Connector Setup

BigQuery (OAuth)

Click “Connect with Google” in the connections dashboard. You'll be redirected to Google's consent screen. No API keys or service accounts needed. Tokens refresh automatically.

Snowflake

Provide your account identifier (e.g. xy12345.us-east-1), username, and password. Specify a warehouse, database, and schema.

Databricks

Generate a personal access token in Databricks: User Settings > Developer > Access Tokens. Provide the token with your workspace URL and target catalog/schema.

Supabase

Find connection details in your Supabase project dashboard under Settings > Database > Connection String. Use the service_role key for full access.

Examples

Ask in natural language. Your AI selects and chains tools automatically.

Basic Profiling

"What's the ADMET profile of ibuprofen? Is it hepatotoxic?"

Response includes properties, ADMET predictions, and compliance status:

{
  "properties": {
    "cid": 3672,
    "smiles": "CC(C)CC1=CC=C(C=C1)C(C)C(=O)O",
    "molecular_weight": 206.28,
    "logp": 3.97,
    "tpsa": 37.3,
    "qed": 0.72
  },
  "admet": {
    "hepatotoxicity_probability": 0.12,
    "cardiotoxicity_probability": 0.08,
    "bbb_permeant": true,
    "cyp2d6_inhibitor": false,
    "human_intestinal_absorption": 0.98
  },
  "compliance": {
    "status": "clear",
    "is_dea_controlled": false,
    "is_fda_banned": false,
    "structural_alerts": []
  }
}

More Examples

"Optimize this lead compound for better oral bioavailability while maintaining selectivity: CC1=CC=C(C=C1)C(=O)NC2=CC=CC=C2"

→ optimize_molecule — generates variants targeting QED and LogP

"Is this molecule safe to develop? Check for any regulatory concerns: CN1C=NC2=C1C(=O)N(C(=O)N2C)C"

→ get_molecule_profile — returns DEA, FDA, structural alert status

"What's known about CDK4/6 inhibitors in breast cancer? Show me literature and clinical trials."

→ search_literature + search_clinical_trials — parallel search

"Find molecules similar to palbociclib with QED > 0.6 and no PAINS alerts"

→ search_similar — similarity search with property and compliance filters

Novel molecules: Compounds not in the pre-computed database are processed on-the-fly via RDKit with FAVES screening. Response times: ~500ms vs ~50ms for indexed compounds.