Quick start

After installing citracer, run your first trace:

From a local PDF

citracer --pdf paper.pdf --keyword "channel-independent" --depth 3

From an arXiv ID

citracer --arxiv 2211.14730 --keyword "self-attention"

From a DOI or URL

citracer --doi 10.48550/arxiv.2211.14730 --keyword "patching"
citracer --url https://openreview.net/forum?id=cGDAkQo1C0p --keyword "instance normalization"

Multiple keywords

citracer --pdf paper.pdf --keyword "channel-independent" --keyword "patching"

By default (--match-mode any), a paper is matched if any keyword appears. Use --match-mode all to require every keyword.

What happens

1. The source PDF is parsed with GROBID
2. Every sentence containing the keyword is found, along with the references cited nearby
3. Those references are downloaded (arXiv, Semantic Scholar, OpenReview, Sci-Hub, preprint servers)
4. The process repeats recursively up to --depth levels
5. An interactive HTML graph opens in your browser

Next steps

• Add --semantic to catch conceptual matches beyond literal keywords: Semantic matching
• Trace who cites a paper instead of what it cites: Reverse trace
• Compare traces over time: Literature monitoring
• See all available flags: CLI reference