Deepxiv Search

deepxiv search "<query>" --date-from <YYYY-MM-DD> --limit <N> --format json

• --date-from: Filter papers by date (e.g., 2026-03-01 for last month)
• --limit: Number of results (default 20)
• --format json: Get JSON output for programmatic use

2. Preview Paper (Low-cost)

deepxiv paper <arxiv_id> --brief

Returns: title, TLDR, keywords, authors, GitHub link, citations. Use this for initial filtering before deep reading.

3. View Paper Structure

deepxiv paper <arxiv_id> --head

See all sections/chapters with token counts. Helps decide which parts to read.

4. Read Specific Section

deepxiv paper <arxiv_id> --section <SectionName>

Read only the specified section (e.g., "Experiments", "Introduction", "Method", "Results").

5. Trending Papers

deepxiv trending --days <N> --limit <N> --json

Get hot papers in the last N days.

6. Web Search (Fallback)

deepxiv wsearch "<query>"

Call internet search for broader results.

7. Semantic Scholar Metadata

deepxiv sc <paper_id>

Get metadata from Semantic Scholar.

8. PMC Papers

deepxiv pmc <pmc_id> --head    # View structure
deepxiv pmc <pmc_id>           # View full text

9. Deep Research Agent

deepxiv agent query "<research_question>" --verbose

Conduct in-depth research on a topic end-to-end.

10. Check Paper Popularity

deepxiv paper <arxiv_id> --popularity

View social media buzz and dissemination metrics.

Typical Workflows

Research into a New Topic

1. Search for candidate papers:

   deepxiv search "<topic>" --date-from <YYYY-MM-DD> --limit 50 --format json
   

2. Preview papers to filter:

   deepxiv paper <id> --brief
   

3. View structure:

   deepxiv paper <id> --head
   

4. Read key sections:

   deepxiv paper <id> --section Experiments
   

5. Compile results into markdown table

Track Recent Trends

1. Get recent papers:

   deepxiv trending --days 30 --limit 50 --json
   

2. Preview relevant ones:

   deepxiv paper <id> --brief
   

3. Check popularity:

   deepxiv paper <id> --popularity
   

Compile a Baseline Table

When user asks "what papers exist on X, what datasets they use, how they perform":

1. Search with date filter
2. Preview with --brief for initial filtering
3. Read --section Experiments for benchmark details
4. Extract: paper title, arXiv link, open-source (Y/N), code link, datasets, metrics, scores
5. Format as markdown table

Output Format Examples

--brief Output

📄 MMA: Multimodal Memory Agent
🆔 arXiv: 2602.16493
📅 Published: 2026-02-18
📊 Citations: 0
🔗 PDF: https://arxiv.org/pdf/2602.16493
💻 GitHub: https://github.com/AIGeeksGroup/MMA
🏷️  Keywords: memory-level reliability, temporal decay, conflict-aware consensus
💡 TLDR: [research paper] MMA introduces a memory-level reliability framework...

--head Output (JSON)

{
  "sections": [
    {"name": "Introduction", "idx": 0, "tldr": "...", "token_count": 1098},
    {"name": "Method", "idx": 1, "tldr": "...", "token_count": 5420},
    {"name": "Experiments", "idx": 2, "tldr": "...", "token_count": 3200}
  ]
}

ArXiv ID Format

• Format: YYMMDD.NNNNN (e.g., 2603.16496 means March 16, 2026, ID 16496)
• PMC IDs: PMCNNNNNN (e.g., PMC544940)

Coverage

• Full arXiv coverage with daily updates
• Expanding to PMC, ACM, bioRxiv/medRxiv/ChemRxiv, Semantic Scholar
• Target: 2B+ open academic papers

Notes

• Use --brief first to save tokens - preview before deep reading
• Use --head to understand structure before deciding what to read
• --section allows targeted reading without consuming full paper
• JSON output enables programmatic processing and table generation
• Combine multiple near-synonym searches to increase recall37:["$","$L3f",null,{"content":"$40","frontMatter":{"name":"deepxiv-search","description":"This skill should be used when the user asks to "search papers", "find academic papers", "look up research papers", "track paper trends", "get paper summaries", "find papers on [topic]", "search arXiv", "academic literature search", "research paper lookup", or needs to find, analyze, or summarize academic papers. Also use when user mentions "DeepXiv", "arXiv search", "paper research", "literature review", or "find recent papers about".","version":"1.0.0"}}]