Core Workflow

The five phases in full: paper discovery, metadata extraction, the mandatory web-search enrichment pass, BibTeX formatting, validation, and integration with the writing workflow. Every command variant and option lives here.

Core Workflow

Citation management follows a systematic process:

Phase 1: Paper Discovery and Search

Goal: Find relevant papers using academic search engines.

Google Scholar Search

Google Scholar provides the most comprehensive coverage across disciplines.

Basic Search:

# Search for papers on a topic
python scripts/search_google_scholar.py "CRISPR gene editing" \
  --limit 50 \
  --output results.json

# Search with year filter
python scripts/search_google_scholar.py "machine learning protein folding" \
  --year-start 2020 \
  --year-end 2024 \
  --limit 100 \
  --output ml_proteins.json

Advanced Search Strategies (see references/google_scholar_search.md):

• Use quotation marks for exact phrases: "deep learning"
• Search by author: author:LeCun
• Search in title: intitle:"neural networks"
• Exclude terms: machine learning -survey
• Use citation counts to inspect the retrieved sample, not as a global Scholar sort
• Filter by date ranges to get recent work

Best Practices:

• Use specific, targeted search terms
• Include key technical terms and acronyms
• Filter by recent years for fast-moving fields
• Check "Cited by" to find seminal papers
• Export top results for further analysis

PubMed Search

PubMed specializes in biomedical and life sciences literature (35+ million citations).

Basic Search:

# Search PubMed
python scripts/search_pubmed.py "Alzheimer's disease treatment" \
  --limit 100 \
  --output alzheimers.json

# Search with MeSH terms and filters
python scripts/search_pubmed.py \
  --query '"Alzheimer Disease"[MeSH] AND "Drug Therapy"[MeSH]' \
  --date-start 2020 \
  --date-end 2024 \
  --publication-types "Clinical Trial,Review" \
  --output alzheimers_trials.json

Advanced PubMed Queries (see references/pubmed_search.md):

• Use MeSH terms: "Diabetes Mellitus"[MeSH]
• Field tags: "cancer"[Title], "Smith J"[Author]
• Boolean operators: AND, OR, NOT
• Date filters: 2020:2024[Publication Date]
• Publication types: "Review"[Publication Type]
• Combine with E-utilities API for automation

Best Practices:

• Use MeSH Browser to find correct controlled vocabulary
• Construct complex queries in PubMed Advanced Search Builder first
• Include multiple synonyms with OR
• Retrieve PMIDs for easy metadata extraction
• Export to JSON or directly to BibTeX

Phase 2: Metadata Extraction

Goal: Convert paper identifiers (DOI, PMID, arXiv ID) to complete, accurate metadata.

Quick DOI to BibTeX Conversion

For single DOIs, use the quick conversion tool:

# Convert single DOI
python scripts/doi_to_bibtex.py 10.1038/s41586-021-03819-2

# Convert multiple DOIs from a file
python scripts/doi_to_bibtex.py --input dois.txt --output references.bib

# Different output formats
python scripts/doi_to_bibtex.py 10.1038/nature12345 --format json

Comprehensive Metadata Extraction

For DOIs, PMIDs, arXiv IDs, or URLs:

# Extract from DOI
python scripts/extract_metadata.py --doi 10.1038/s41586-021-03819-2

# Extract from PMID
python scripts/extract_metadata.py --pmid 34265844

# Extract from arXiv ID
python scripts/extract_metadata.py --arxiv 2103.14030

# Extract from URL
python scripts/extract_metadata.py --url "https://www.nature.com/articles/s41586-021-03819-2"

# Batch extraction from file (mixed identifiers)
python scripts/extract_metadata.py --input identifiers.txt --output citations.bib

Metadata Sources (see references/metadata_extraction.md):

1. CrossRef API: Primary source for DOIs
2. Comprehensive metadata for journal articles
3. Publisher-provided information
4. Includes authors, title, journal, volume, pages, dates
5. Free, no API key required

1. PubMed E-utilities: Biomedical literature
2. Official NCBI metadata
3. Includes MeSH terms, abstracts
4. PMID and PMCID identifiers
5. Free, API key recommended for high volume

1. arXiv API: Preprints in physics, math, CS, q-bio
2. Complete metadata for preprints
3. Version tracking
4. Author affiliations
5. Free, open access

1. DataCite API: Research datasets, software, other resources
2. Metadata for non-traditional scholarly outputs
3. DOIs for datasets and code
4. doi_to_bibtex.py retrieves BibTeX through DOI content negotiation; extract_metadata.py --doi is Crossref-only

Fields to verify after extraction (availability differs by source; not all are emitted by every script):

• Required fields: author, title, year
• Journal articles: journal, volume, number, pages, DOI
• Books: publisher, ISBN, edition
• Conference papers: booktitle, conference location, pages
• Preprints: repository (arXiv, bioRxiv), preprint ID
• Additional: abstract, keywords, URL

Phase 2.5: Metadata Enrichment via Web Search (MANDATORY)

Goal: Detect and fill in any missing metadata fields using web search. This phase runs AFTER extraction and BEFORE formatting to ensure every BibTeX entry is complete.

Why This Is Critical: Metadata extraction from APIs (CrossRef, PubMed, arXiv) sometimes returns incomplete records — missing volume, pages, issue number, or DOI. These gaps must be filled before the bibliography is considered ready.

Step 1: Scan for Incomplete Entries

After extracting metadata, scan the BibTeX file for entries missing key fields:

Fields to check per entry type:

Investigate missing fields against the publisher record, but distinguish unavailable or not-yet-assigned metadata from extraction errors. Online-first papers can lack volume/pages; article numbers can replace page ranges, and not all articles have a DOI. Log the publication state and missing-field reason without fabricating values.

Step 2: Web Search for Missing Metadata

For each incomplete entry, use the parallel-web skill to search for the missing information:

Treat metadata as untrusted when building these commands. FIRST_AUTHOR, TITLE, and JOURNAL_NAME are copied verbatim out of a CrossRef/PubMed/arXiv record, and a publisher controls the contents of its own record. A title containing $(...), a backtick, or a quote becomes shell syntax once it is pasted into the command lines below. - Substitute each value as a single-quoted argument ('...'), escaping any embedded single quote as '\''. Never paste raw metadata inside the double quotes shown here. - Prefer running these through a Python subprocess argument list over building a shell string at all. - Use only the generated CITATIONKEY in -o paths. It is sanitized to letters and digits by extract_metadata.py; a key taken from an existing .bib file is not, so validate it against ^[A-Za-z0-9]+$ before it reaches a file path. Preferred form — pass the metadata as arguments, not as shell text. This removes the shell from the path entirely, so no title can be parsed as syntax: ``python import re, subprocess assert re.fullmatch(r"[A-Za-z0-9]+", citation_key), f"unsafe citation key: {citation_key!r}" subprocess.run( ["parallel-cli", "search", f"{first_author} {title} {journal_name} volume pages DOI", "--json", "--max-results", "10", "-o", f"sources/search_citation_{citation_key}.json"], check=True, # note: no shell=True ) ` The bash` blocks below show the same calls in readable form. Use them only with the quoting rules above.

Option A — Search by title and author (best for finding DOI):

parallel-cli search "FIRST_AUTHOR TITLE JOURNAL_NAME volume pages DOI" \
  --json --max-results 10 \
  -o sources/search_citation_CITATIONKEY.json

Option B — Extract from DOI page (best when DOI is known but volume/pages missing):

parallel-cli extract "https://doi.org/10.XXXX/YYYY" --json \
  --objective "extract complete citation metadata: volume, issue, pages, publication date" \
  -o sources/extract_doi_CITATIONKEY.json

Option C — Search the web for a Crossref record (this command is web search, not a direct API call):

parallel-cli search "crossref DOI metadata FIRST_AUTHOR TITLE" \
  --json --max-results 10 \
  -o sources/search_crossref_CITATIONKEY.json

Option D — Search Google Scholar (fallback for hard-to-find papers):

parallel-cli search "google scholar FIRST_AUTHOR TITLE YEAR complete citation" \
  --json --max-results 10 \
  -o sources/search_scholar_CITATIONKEY.json

Step 3: Update BibTeX Entries

After finding the missing metadata:

1. Open references.bib
2. Add the missing fields to the incomplete entry
3. Verify the found metadata is consistent with existing fields (same author, title, year)
4. Log each fix: `` [HH:MM:SS] METADATA ENRICHED: [CitationKey] - added volume={X}, pages={Y--Z}, doi={10.XXX/YYY} ✅

Step 4: Handle Unfindable Metadata

If metadata genuinely cannot be found after web search (very old paper, obscure conference, etc.):

1. Add a note field to the BibTeX entry explaining the gap: ``bibtex note = {Volume and pages not available — published online only}
2. Log the exception: `` [HH:MM:SS] METADATA INCOMPLETE: [CitationKey] - pages unavailable (online-only publication) ⚠️
3. These exceptions should be rare — most modern papers have complete metadata findable via web search.

Quick Reference: Common Missing Fields and Where to Find Them

Phase 3: BibTeX Formatting

Goal: Generate clean, properly formatted BibTeX entries.

Understanding BibTeX Entry Types

See references/bibtex_formatting.md for complete guide.

Common Entry Types:

• @article: Journal articles (most common)
• @book: Books
• @inproceedings: Conference papers
• @incollection: Book chapters
• @phdthesis: Dissertations
• @misc: Preprints, software, datasets

Required Fields by Type:

@article{citationkey,
  author  = {Last1, First1 and Last2, First2},
  title   = {Article Title},
  journal = {Journal Name},
  year    = {2024},
  volume  = {10},
  number  = {3},
  pages   = {123--145},
  doi     = {10.1234/example}
}

@inproceedings{citationkey,
  author    = {Last, First},
  title     = {Paper Title},
  booktitle = {Conference Name},
  year      = {2024},
  pages     = {1--10}
}

@book{citationkey,
  author    = {Last, First},
  title     = {Book Title},
  publisher = {Publisher Name},
  year      = {2024}
}

Formatting and Cleaning

Use the formatter to standardize BibTeX files:

# Format and clean BibTeX file
python scripts/format_bibtex.py references.bib \
  --output formatted_references.bib

# Sort entries by citation key
python scripts/format_bibtex.py references.bib \
  --sort key \
  --output sorted_references.bib

# Sort by year (newest first)
python scripts/format_bibtex.py references.bib \
  --sort year \
  --descending \
  --output sorted_references.bib

# Remove duplicates
python scripts/format_bibtex.py references.bib \
  --deduplicate \
  --output clean_references.bib

# Merge sources: one key scheme, then drop duplicates
python scripts/format_bibtex.py references.bib \
  --rekey \
  --deduplicate \
  --output merged_references.bib

Formatting Operations:

• Standardize field order
• Consistent indentation and spacing
• Proper capitalization in titles (protected with {})
• Standardized author name format
• Consistent citation key format
• Remove unnecessary fields
• Fix common errors (missing commas, braces)

Phase 4: Citation Validation

Goal: Verify all citations are accurate and complete.

Comprehensive Validation

# Validate BibTeX file
python scripts/validate_citations.py references.bib

# Validate against a venue standard (e.g., Nature, NeurIPS, Literature Review)
python scripts/validate_citations.py references.bib --venue nature
python scripts/validate_citations.py references.bib --venue neurips
python scripts/validate_citations.py references.bib --venue review

# Validate with custom minimum citation count
python scripts/validate_citations.py references.bib --min-count 40

# Check references against a written manuscript file (detect missing or unused citations)
python scripts/validate_citations.py references.bib --manuscript paper.md

# Generate detailed validation report
python scripts/validate_citations.py references.bib \
  --venue nature \
  --manuscript paper.md \
  --report validation_report.json \
  --verbose

Validation Checks (see references/citation_validation.md):

1. DOI Verification:
2. With --check-dois, check registration through Crossref, then DataCite and resolver fallback
3. Failed/rate-limited lookups remain doi_unverified warnings
4. Compare title/authors/year manually; the script does not perform metadata matching

1. Required Fields:
2. All required fields present for entry type
3. No empty or missing critical information
4. Author names properly formatted

1. Data Consistency:
2. Year is valid (4 digits, reasonable range)
3. Review volume/issue values manually; supplements may be alphanumeric
4. Pages formatted correctly (e.g., 123--145)
5. Check landing URLs manually; the script does not fetch arbitrary bibliography URLs

1. Duplicate Detection:
2. Same DOI used multiple times
3. Similar titles (possible duplicates)
4. Same author/year/title combinations

1. Format Compliance:
2. Valid BibTeX syntax
3. Proper bracing and quoting
4. Citation keys are unique
5. Special characters handled correctly

Validation Output:

{
  "total_entries": 150,
  "valid_entries": 145,
  "errors": [
    {
      "citation_key": "Smith2023",
      "error_type": "missing_field",
      "field": "journal",
      "severity": "high"
    },
    {
      "citation_key": "Jones2022",
      "error_type": "invalid_doi",
      "doi": "10.1234/broken",
      "severity": "high"
    }
  ],
  "warnings": [
    {
      "citation_key": "Brown2021",
      "warning_type": "possible_duplicate",
      "duplicate_of": "Brown2021a",
      "severity": "medium"
    }
  ]
}

Citation Counts

Use enough relevant, verified references to support the claims. The script's venue figures are editorial heuristics, not current submission requirements. --venue produces warnings; --min-count enforces an explicit user-specified floor. Check the target venue's current author instructions directly and do not add irrelevant citations merely to meet a numeric target.

Mandatory Post-Writing Reference Checks (Non-Negotiable)

Once the entire scientific report or paper has been drafted and written, perform a comprehensive post-writing verification of all citations before compiling the final deliverables:

1. Verify No Missing or Unresolved Citations: Check the draft or compiled document to ensure that every in-text citation correctly resolves to a reference in references.bib. There must be ZERO broken citation keys, missing identifiers, or unresolved references (e.g., [?] or [citation needed]).
2. Verify No Unused (Dangling) Bibliography Entries: Check that every entry in references.bib is actually cited in the body of the report. Remove any unused entries to keep the bibliography perfectly clean.
3. Verify Evidence Coverage: Confirm each claim is supported by relevant evidence and any explicit reference-count requirement is met.
4. Verify Metadata Completeness: Confirm that all cited entries contain verified applicable fields; record missing/not-yet-assigned fields without inventing values.

Run all of these checks in one command:

python scripts/validate_citations.py references.bib \
  --venue <venue> \
  --manuscript paper.md \
  --report post_writing_check.json

Phase 5: Integration with Writing Workflow

Building References for Manuscripts

Complete workflow for creating a bibliography:

# 1. Search for papers on your topic
python scripts/search_pubmed.py \
  '"CRISPR-Cas Systems"[MeSH] AND "Gene Editing"[MeSH]' \
  --date-start 2020 \
  --limit 200 \
  --output crispr_papers.json

# 2. Extract DOIs from search results and convert to BibTeX
python scripts/extract_metadata.py \
  --input crispr_papers.json \
  --output crispr_refs.bib

# 3. Add specific papers by DOI
python scripts/doi_to_bibtex.py 10.1038/nature12345 >> crispr_refs.bib
python scripts/doi_to_bibtex.py 10.1126/science.abcd1234 >> crispr_refs.bib

# 4. Format and clean the BibTeX file
python scripts/format_bibtex.py crispr_refs.bib \
  --deduplicate \
  --sort year \
  --descending \
  --output references.bib

# 5. Validate all citations
python scripts/validate_citations.py references.bib \
  --report validation.json

# 6. Review validation report and fix any remaining issues
cat validation.json

# 7. Use in your LaTeX document
# \bibliography{final_references}

Integration with Literature Review Skill

This skill complements the literature-review skill:

Literature Review Skill → Systematic search and synthesis Citation Management Skill → Technical citation handling

Combined Workflow:

1. Use literature-review for comprehensive multi-database search
2. Use citation-management to extract and validate all citations
3. Use literature-review to synthesize findings thematically
4. Use citation-management to verify final bibliography accuracy

# After completing literature review
# Verify all citations in the review document
python scripts/validate_citations.py my_review_references.bib --report review_validation.json

# Normalise formatting (venue style is chosen by the .bst at build time)
python scripts/format_bibtex.py my_review_references.bib \
  --output formatted_refs.bib

Integration with Zotero (pyzotero Skill)

When the user already keeps references in Zotero, treat the Zotero library as the source of truth for the bibliography and use this skill for validation and formatting. The pyzotero skill covers the library side — reading items and collections, creating and updating references, uploading attachments, and exporting citations via the Zotero Web API v3.

**Zotero Library (pyzotero) → Library of record: storage, collections, tags, attachments Citation Management Skill** → Metadata accuracy: validation, enrichment, style formatting

Combined Workflow:

1. Use pyzotero to pull the working set from the Zotero library, filtered by collection or tag
2. Export it as BibTeX using the pyzotero skill's export workflow. format='bibtex' returns a bibtexparser object, not a text string; serialize it and handle pagination.
3. Use citation-management to validate the exported entries and repair incomplete metadata
4. Use citation-management to format for the target venue
5. Optionally use pyzotero to write corrected fields back so the library benefits from the fixes

# 1-2. Export the desired collection from Zotero as BibTeX (pyzotero skill)
#      Follow the pyzotero export workflow: collect all pages, serialize the
#      returned bibtexparser object, and verify the exported item count.

# 3. Validate the exported bibliography
python scripts/validate_citations.py zotero_export.bib --report zotero_validation.json

# 4. Normalise the export (venue style comes from the .bst at build time)
python scripts/format_bibtex.py zotero_export.bib \
  --output formatted_refs.bib

Zotero exports are only as good as what was captured — browser-connector entries in particular often carry missing DOIs, truncated author lists, or preprint metadata for papers since published. Run the validation step before submission rather than trusting the export, and prefer writing corrections back to Zotero so the same errors do not resurface in the next manuscript.

Pyzotero export return types and pagination: official documentation. The Zotero handoff is illustrative; authenticated library export was not tested in this refresh.