| name | pubchem-search |
| description | Query PubChem by name, CID, or SMILES; retrieve properties; run similarity/substructure searches; and get bioactivity. Use for cheminformatics on a specific chemical, drug, or molecule. Public PUG-REST API, no credential needed. |
| license | Apache-2.0 |
Pubchem Search (Codex-native)
Codex-native adaptation of Google DeepMind's pubchem-database skill
(science-skills, Apache-2.0).
Uses the public API directly through Codex's own tools.
Gates (ask before proceeding)
- Terms notice: on first use in a workspace, tell the user to review the
source's terms of use and record the notice in provenance.
- Network / install: read-only public access; be polite about rate limits.
Write large responses to a file and parse them; if you write helper code, run
it via
uv run, never bare python3. Ask before installs or large downloads.
Endpoints
- PubChem PUG-REST
https://pubchem.ncbi.nlm.nih.gov/rest/pug
Workflow
Resolve compound names and common properties first with
science_search_pubchem; use PUG REST directly for structure search and assays.
- Resolve name/SMILES to a CID (
/compound/name/<name>/cids/JSON), then fetch properties (/compound/cid/<cid>/property/<props>/JSON).
- For similarity/substructure, use the
/compound/fastsimilarity_2d or /fastsubstructure endpoints.
- Provenance & review: record queries, endpoints, and cited identifiers with
$science-provenance; check claims with $science-review before presenting.
Boundaries
- Report the CID and canonical/isomeric SMILES; a name can map to multiple CIDs (salts, stereoisomers).
- Cite only CIDs actually returned; never invent structures or properties.
