Common task recipes

Install the Python or JavaScript SDK and configure your API key first. These examples use Python; the corresponding JavaScript methods and REST routes are listed below. Each chemistry or corpus operation consumes one successful metered call.

Find how a molecule was preparedCopy Anchor Link

Search the product structure, then retrieve the full source for a selected result:

from cyberrxn import Client

with Client() as client:
    matches = client.molecule_reactions(
        "CCN1CCN(CO)C(=O)C1=O", role="product", limit=3
    )
    if matches.items:
        record = client.get_observation(matches.items[0].observation_id)
        print(record.evidence.patent, record.evidence.year)
        print(record.paragraph_text)
        print(record.conditions)
    else:
        print("No recorded match for this structure.")

A returned hit is evidence to inspect, not a recommended procedure. Pressure, solvent and temperature can belong to different stages. Missing fields remain unknown. Search semantics.

Search an exact transformationCopy Anchor Link

Use lookup_reaction when both sides are known; use search_reactions when you need filters:

from cyberrxn import Client

with Client() as client:
    exact = client.lookup_reaction(
        "C=O.CCN1CCNC(=O)C1=O>>CCN1CCN(CO)C(=O)C1=O", limit=3
    )
    filtered = client.search_reactions(
        products="CCN1CCN(CO)C(=O)C1=O",
        temperature_min=20, temperature_max=80, limit=3,
    )
    print("Exact observations:", len(exact.items))
    print("Filtered observations:", len(filtered.items))

These are two alternative searches, shown together for comparison: running the whole snippet makes two calls. Agents are not part of the identity of an exact transformation. Structural analogues from similar_products have a different meaning; similarity is not synthesis probability.

Check, balance and display an equationCopy Anchor Link

from cyberrxn import Client

with Client() as client:
    check = client.validate_reaction("CCO>>CC=O")
    print(check.valid, check.atom_balance.delta)
    balance = client.balance_reaction(
        "[H][H].O=O>>O", method="coefficients"
    )
    if balance.solved and balance.balanced_reaction_smiles:
        display = client.convert_smiles(balance.balanced_reaction_smiles)
        print(display.unicode)

This performs two calls, plus a third if conversion runs. The coefficient solver balances supplied species only. method="synrbl" is a separate, explicit model-assisted hypothesis for missing species; inspect imputed_species and limitations. Neither computation establishes experimental evidence. Validation and balance.

Convert a structure for another toolCopy Anchor Link

from cyberrxn import Client

with Client() as client:
    result = client.convert_representation(
        "CCO", input_format="smiles", output_format="inchi"
    )
    print(result)

Supply explicit input and output formats. Preserve multiline MOL/SDF/RXN data. Conversions that discard information require allow_lossy=True; inspect preservation diagnostics rather than assuming a round trip is lossless. A formula or InChIKey cannot identify a unique structure. Conversion guide.

Equivalent interfacesCopy Anchor Link

For an assistant, ask the same task in plain language after connecting MCP. For a batch, apply rate controls, stop pagination deliberately and inspect usage.