pubchatter 0.1.0

Citation metadata resolver, formatter, verifier, and manuscript citation audit CLI.

PubChatter

PubChatter is a standalone citation-verification CLI and drop-in agent plugin for resolving scholarly metadata from real sources before citations are written into manuscripts.

It uses public metadata APIs and does not require API keys:

• CrossRef for DOI resolution and general scholarly search
• PubMed E-utilities for biomedical PMID lookup and PubMed search
• Semantic Scholar for citation counts, abstracts, and CS/bio search
• OpenAlex for DOI fallback and open scholarly metadata

Capabilities

• Resolve DOI metadata into structured JSON or formatted citations.
• Resolve PubMed IDs.
• Search across CrossRef, PubMed, and Semantic Scholar.
• Search one source directly with pubmed, crossref, or s2.
• Verify a free-form citation string against real metadata.
• Batch-resolve DOI/PMID lists.
• Format references in nature, apa, vancouver, bibtex, cell, or science style.
• Audit .docx manuscripts for broken inline citation superscripts.
• Auto-fix known broken .docx citation superscripts for common method names.
• Report DOI coverage in manuscript reference lists.

Standalone Install

From PyPI:

python -m pip install pubchatter

From GitHub:

python -m pip install "git+https://github.com/nanogenomic/pubchatter.git"

From a local checkout:

python -m pip install -e .

Then run:

pubchatter --help

python-docx is declared as a dependency so the .docx audit commands work after installation.

Plugin-Style Install

PubChatter can also be used as a drop-in local plugin by placing this directory at:

~/.claude/plugins/pubchatter

For a local executable wrapper:

mkdir -p ~/.local/bin
ln -sf ~/.claude/plugins/pubchatter/cli.py ~/.local/bin/pubchatter
chmod +x ~/.claude/plugins/pubchatter/cli.py

If you prefer the wrapper script used in the original local setup:

#!/usr/bin/env bash
exec python3 ~/.claude/plugins/pubchatter/cli.py "$@"

Examples

Resolve a DOI:

pubchatter doi 10.1038/s41586-021-03819-2

Return JSON:

pubchatter doi 10.1038/s41586-021-03819-2 --json

Format a citation:

pubchatter format 10.1038/s41586-021-03819-2 --style nature

Search broadly:

pubchatter search "peptide binder design deep learning" --max-results 5

Search one source:

pubchatter pubmed "protein protein binding affinity"
pubchatter crossref "AlphaFold protein structure prediction"
pubchatter s2 "discrete diffusion language model"

Verify a citation:

pubchatter verify "Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583-589 (2021). https://doi.org/10.1038/s41586-021-03819-2"

Batch-resolve identifiers:

pubchatter batch dois.txt --style science

Audit a manuscript:

pubchatter check-super manuscript.docx
pubchatter check-super manuscript.docx --json

Fix known broken superscript citations in place:

pubchatter fix-super manuscript.docx

Write the fixed copy to a new file:

pubchatter fix-super manuscript.docx --output manuscript.citation-fixed.docx

Manuscript Citation Workflow

Recommended agent workflow:

1. Resolve every known DOI with pubchatter doi <doi>.
2. For uncertain references, use pubchatter search, pubchatter pubmed, pubchatter crossref, or pubchatter s2.
3. Verify free-form references with pubchatter verify "<citation>".
4. Run pubchatter check-super <file.docx> before submission.
5. Use pubchatter fix-super <file.docx> --output <fixed.docx> only after reviewing the reported changes.

Notes

• PubChatter is metadata-first. It should flag uncertainty instead of inventing citation details.
• Some canonical scholarly records do not have DOI metadata, especially OpenReview pages, GitHub software references, and selected arXiv or conference pages. Use their canonical URLs when a DOI is unavailable.
• Citation verification is search-based for free-form strings; always inspect warnings about keyword matches or mismatched authors/years.

License

MIT. See LICENSE.