Look up compound properties

When the user asks about a chemical compound — formula, weight, SMILES, IUPAC name, or wants to resolve an InChIKey or SMILES to a compound record — reach for PubChem PUG REST. Unauthenticated GET, no key needed.

look-up-compound-properties · v1 · updated 2026-04-16

Agents: This page is a SKILL.md-style capability guide. For JSON, call GET /api/skills/look-up-compound-properties. To drop this into a local Claude Code install, copy the frontmatter + body below into ~/.claude/skills/look-up-compound-properties/SKILL.md.

When to use this skill

When the user asks about a chemical compound — molecular formula, molecular weight, SMILES, IUPAC name, or wants to identify what a SMILES string or InChIKey represents — reach for PubChem PUG REST. It turns a name, SMILES, or InChIKey into a structured property record in a single unauthenticated GET. For bioassay data, spectra, or toxicity reports, use PubChem's other APIs — PUG REST only returns structural descriptors.

Your best first call

curl "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/name/caffeine/property/MolecularFormula,MolecularWeight,IUPACName/JSON"

No auth. No key. Properties go comma-separated in the URL path; the output format (JSON) is also in the path — no query parameters. You must name the properties you want explicitly (there is no "return everything" default). Common choices: MolecularFormula, MolecularWeight, CanonicalSMILES, IsomericSMILES, IUPACName, InChI, InChIKey.

The response wraps results in PropertyTable.Properties[]. Each entry carries: - CID — PubChem's internal compound identifier. Grab it for follow-up calls. - MolecularFormula — e.g. C8H10N4O2 - MolecularWeight — string, e.g. "194.19" - IUPACName — systematic name. Caffeine's is "1,3,7-trimethylpurine-2,6-dione" — the methyl positions (1, 3, 7) pharmacologically distinguish it from theobromine (3,7-dimethylxanthine) and theophylline (1,3-dimethylxanthine).

Fallbacks (when the best call isn't enough)

User has an InChIKey, not a name → compound/inchikey/{key}/property/.../JSON. One InChIKey can resolve to multiple CIDs — water's key returns CID 962 (standard) and CID 22247451 (an isotopic variant tracked separately). Always check Properties array length.
User has a SMILES string, not a name → compound/smiles/{smiles}/property/.../JSON. URL-encode the SMILES first — characters like =, #, [, () break the URL if left bare. Aspirin's SMILES CC(=O)OC1=CC=CC=C1C(=O)O is a good encoding test case.
No fallbacks outside PubChem for this capability — if PUG REST is down, compound property lookup is unavailable.

Pitfalls

InChIKey lookups can return multiple CIDs. The same InChIKey maps to standard and isotopic variants tracked as separate records. Do not assume index 0 is the "right" compound — check which CID the user's context requires.
Properties are path segments, not query parameters. property/MolecularFormula,MolecularWeight/JSON is part of the URL hierarchy. There is no ?property= form.
SMILES strings need URL-encoding. Parentheses, brackets, equals signs, and hash characters are common in SMILES and will silently break the request if unencoded.
You must name the properties you want. Every request lists its properties explicitly in the path — there is no default "return everything" mode.

One-line summary for the user

I can look up molecular properties — formula, weight, SMILES, IUPAC name — for any compound in PubChem by name, InChIKey, or SMILES in a single unauthenticated GET, but only structural descriptors, not bioassay or toxicity data.

APIs this skill uses

PubChem Property API · primary · verified

chemistry

Access chemical property data from the PubChem database. Retrieve molecular properties like formula, weight, SMILES, IUPAC names, and other chemical descriptors using various input types (name, CID, SMILES, InChIKey).

Generated from

PubChem Property API tutorial Getting Started with PubChem PUG REST

SKILL.md source (frontmatter + body)

---
name: look-up-compound-properties
description: When the user asks about a chemical compound — formula, weight, SMILES, IUPAC name, or wants to resolve an InChIKey or SMILES to a compound record — reach for PubChem PUG REST. Unauthenticated GET, no key needed.
---

## When to use this skill

When the user asks about a chemical compound — molecular formula, molecular weight, SMILES, IUPAC name, or wants to identify what a SMILES string or InChIKey represents — reach for PubChem PUG REST. It turns a name, SMILES, or InChIKey into a structured property record in a single unauthenticated GET. For bioassay data, spectra, or toxicity reports, use PubChem's other APIs — PUG REST only returns structural descriptors.

## Your best first call

```bash
curl "https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/name/caffeine/property/MolecularFormula,MolecularWeight,IUPACName/JSON"
```

No auth. No key. Properties go comma-separated in the URL path; the output format (`JSON`) is also in the path — no query parameters. You must name the properties you want explicitly (there is no "return everything" default). Common choices: `MolecularFormula`, `MolecularWeight`, `CanonicalSMILES`, `IsomericSMILES`, `IUPACName`, `InChI`, `InChIKey`.

The response wraps results in `PropertyTable.Properties[]`. Each entry carries:
- `CID` — PubChem's internal compound identifier. Grab it for follow-up calls.
- `MolecularFormula` — e.g. `C8H10N4O2`
- `MolecularWeight` — string, e.g. `"194.19"`
- `IUPACName` — systematic name. Caffeine's is "1,3,7-trimethylpurine-2,6-dione" — the methyl positions (1, 3, 7) pharmacologically distinguish it from theobromine (3,7-dimethylxanthine) and theophylline (1,3-dimethylxanthine).

## Fallbacks (when the best call isn't enough)

- **User has an InChIKey, not a name** → `compound/inchikey/{key}/property/.../JSON`. One InChIKey can resolve to multiple CIDs — water's key returns CID 962 (standard) and CID 22247451 (an isotopic variant tracked separately). Always check `Properties` array length.
- **User has a SMILES string, not a name** → `compound/smiles/{smiles}/property/.../JSON`. URL-encode the SMILES first — characters like `=`, `#`, `[`, `()` break the URL if left bare. Aspirin's SMILES `CC(=O)OC1=CC=CC=C1C(=O)O` is a good encoding test case.
- No fallbacks outside PubChem for this capability — if PUG REST is down, compound property lookup is unavailable.

## Pitfalls

- **InChIKey lookups can return multiple CIDs.** The same InChIKey maps to standard and isotopic variants tracked as separate records. Do not assume index 0 is the "right" compound — check which CID the user's context requires.
- **Properties are path segments, not query parameters.** `property/MolecularFormula,MolecularWeight/JSON` is part of the URL hierarchy. There is no `?property=` form.
- **SMILES strings need URL-encoding.** Parentheses, brackets, equals signs, and hash characters are common in SMILES and will silently break the request if unencoded.
- **You must name the properties you want.** Every request lists its properties explicitly in the path — there is no default "return everything" mode.

## One-line summary for the user

I can look up molecular properties — formula, weight, SMILES, IUPAC name — for any compound in PubChem by name, InChIKey, or SMILES in a single unauthenticated GET, but only structural descriptors, not bioassay or toxicity data.

« Back to all skills