Tell us what you want to do. We'll recommend the tools, models, agents, and setup you need.
Use the documented Boltz-2 YAML interface to predict a protein–ligand complex, inspect confidence and optional affinity files, and keep their scientific interpretations separate.
Boltz
Protein & Drug Research
Python
View Setup →Validate and canonicalize SMILES with RDKit without dropping original rows; flag duplicates and optionally enrich valid entries with PubChem identifiers.
RDKit + PubChemPy
Chemical Data
Python
View Setup →Add a pinned, restricted RDKit MCP server to a Cursor project and verify molecular-analysis tools using reproducible SMILES prompts.
RDKit MCP Server (TandemAI) + RDKit
Build a Chemistry Assistant
Cursor · Python
View Setup →Build a traceable literature workflow for medicinal and biological chemistry: search PubMed, fetch article records, and summarize only the retrieved evidence with identifiers.
PubMed MCP (cyanheads)
Search Literature
Claude Desktop
View Setup →Resolve chemical names to PubChem identifiers and retrieve traceable compound properties from Claude Desktop through a locally installed PubChem MCP server.
PubChem MCP (cyanheads) + PubChem PUG REST
Analyze Molecules
Claude Desktop
View Setup →Calculate molecular properties with RDKit and explain the recorded results using a local Ollama model, through an explicit Python bridge.
RDKit
Run AI Locally
Python · Local AI / Ollama
View Setup →Use the current Materials Project Python API to fetch a small material summary, preserve identifiers and units, and separate computed properties from experimental evidence.
Materials Project API
Materials Research
Python
View Setup →Connect Claude Desktop to a local RDKit MCP server to calculate molecular formula, molecular weights, TPSA and Crippen descriptors from SMILES, with reproducible setup and acceptance checks.
RDKit MCP Server (TandemAI) + RDKit
Analyze Molecules
Claude Desktop · Python
View Setup →Generate a small monomer backbone with RFdiffusion, pass its PDB coordinates to ProteinMPNN, and retain paired structures, sequences and provenance for later validation.
RFdiffusion + ProteinMPNN
Protein & Drug Research
Python
View Setup →Retrieve reviewed UniProtKB protein records through Claude Desktop, disambiguate organism and accession, and retain functional evidence and source identifiers.
UniProt MCP (cyanheads)
Protein & Drug Research
Claude Desktop
View Setup →Combine RCSB structure records with UniProt protein annotations, inspect method/resolution and entity mappings, and download a traceable mmCIF file.
UniProt MCP (cyanheads) + RCSB MCP (cnyambura) + RCSB PDB Data API
Protein & Drug Research
Claude Desktop · Python
View Setup →Prepare QM9 molecular graphs, train a small CPU GCN baseline with training-only normalization, and report validation-selected test MAE in Debye.
PyTorch Geometric
Predict Properties
Python
View Setup →