nmrxiv - SKILL.md Agent Skill

name: nmrxiv description: Search and download NMR spectroscopy datasets from nmrxiv.org. Use when the user asks to find NMR data, search for HSQC/HMBC/COSY/DEPT experiments, look up compounds by name or SMILES, download NMR datasets, or access nmrxiv.org data programmatically. Outputs JSON for easy parsing.

nmrxiv-downloader: NMR Dataset Search & Download

CLI tool for searching and downloading NMR spectroscopy datasets from nmrxiv.org.

Setup

# Install from PyPI
pip install nmrxiv-downloader

# Verify installation
nmrxiv --help

Quick Reference

Task	Command	JSON Path
List projects	`nmrxiv list --type project`	`.items[].identifier`
List datasets	`nmrxiv list --type dataset`	`.items[].identifier`
Search molecule	`nmrxiv search --query "name"`	`.results[].canonical_smiles`
Search SMILES	`nmrxiv search --smiles "CCO"`	`.results[].iupac_name`
Filter by experiment	`nmrxiv search --type hsqc`	`.results[].identifier`
Show details	`nmrxiv show P5`	`.item.download_url`
Download	`nmrxiv download P5 -o /tmp`	`.file`, `.size`
Download + extract	`nmrxiv download P5 -o /tmp --extract`	`.extracted_to`

Common Experiment Types

nmrxiv search --type hsqc      # 1H-13C HSQC
nmrxiv search --type hmbc      # HMBC
nmrxiv search --type cosy      # COSY
nmrxiv search --type dept      # DEPT
nmrxiv search --type "1d-13c"  # 1D 13C
nmrxiv search --type noesy     # NOESY
nmrxiv search --type tocsy     # TOCSY

Key Facts

All output is JSON by default (add --no-json for human-readable)
Projects have download URLs, datasets don't - download via parent project
Data hierarchy: Project → Study → Dataset → Files (Bruker, JCAMP-DX)
Pagination: Use --page N for large result sets

Workflow: Find and Download NMR Data

# 1. Search for experiment type
nmrxiv search --type hsqc

# 2. Get dataset details (shows parent project)
nmrxiv show D410

# 3. Download parent project
nmrxiv download P11 --output ./data --extract

# 4. Data is now in ./data/P11/

Workflow: Search by Compound

# 1. Search by name
nmrxiv search --query "quercetin"

# 2. Or by SMILES substructure
nmrxiv search --smiles "c1ccc(O)cc1"

# 3. Find related projects
nmrxiv list --type project | jq '.items[] | select(.name | test("quercetin"; "i"))'

Error Handling

If download fails with "No download URL":

{"error": true, "message": "No download URL for D410. Try downloading parent project: P11", "code": 1}

→ Use the suggested parent project ID to download.

Parsing Examples

# Get first project identifier
nmrxiv list --type project | jq -r '.items[0].identifier'

# Get all HSQC dataset IDs
nmrxiv search --type hsqc | jq -r '.results[].identifier'

# Get download URL for a project
nmrxiv show P5 | jq -r '.item.download_url'