Examples

Research Assistants

Build AI agents that can search, read, and cite papers accurately with stable node IDs.

Prompt:

"Find papers on attention mechanisms and explain equation 1 from the transformer paper"

Agent workflow:

1. searchPapers(["attention mechanisms"]) → finds 1706.03762
2. getPaperOverview(["1706.03762"]) → discovers eq:1 exists
3. getNodes([{ arxivId: "1706.03762", nodeIds: ["eq:1"] }]) → fetches equation content

Literature Review

Automate literature surveys by traversing citation graphs forward and backward.

Prompt:

"What papers cite the original transformer paper? Summarize their contributions."

Agent workflow:

1. getCitations("1706.03762") → finds BERT, GPT-2, etc.
2. getPaperOverview(["1810.04805", "..."]) → gets abstracts and summaries
3. Agent synthesizes findings into a literature review

RAG Pipelines

Ground LLM responses in verifiable paper content with precise section references.

Workflow:

1. Use searchPapers to find relevant papers
2. Use getPaperOverview to get document structure
3. Chunk papers by section using getNodes
4. Embed chunks with nodeIds for retrieval
5. Cite sources with stable references: "See Section 3.2.1 of [1706.03762]"

IDE Integrations

Add paper lookup to Cursor, VS Code, or any MCP-compatible editor.

Example prompt:

"Explain this loss function in my code"

Agent workflow:

1. Extracts loss function name from code context
2. searchPapers(["cross entropy loss", "softmax loss"]) → finds relevant papers
3. getNodes([...]) → fetches equation and explanation
4. Returns annotated explanation with paper citations

Batch Research

Efficiently gather information across multiple papers.

Prompt:

"Compare the attention mechanisms in these 5 papers"

Agent workflow:

// Single call to get all paper structures
getPaperOverview({
  "arxivIds": ["1706.03762", "1810.04805", "2005.14165", "2103.14030", "2201.08239"]
})

// Single call to fetch specific sections from all papers
getNodes({
  "requests": [
    { "arxivId": "1706.03762", "nodeIds": ["sec:3.2"] },
    { "arxivId": "1810.04805", "nodeIds": ["sec:3.1"] },
    { "arxivId": "2005.14165", "nodeIds": ["sec:2"] },
    { "arxivId": "2103.14030", "nodeIds": ["sec:3"] },
    { "arxivId": "2201.08239", "nodeIds": ["sec:2.1"] }
  ]
})

Citation Graph Exploration

Build understanding of a research field by exploring connections.

Prompt:

"What are the foundational papers in transformer research?"

Agent workflow:

1. Start with known seminal paper: 1706.03762
2. getReferences("1706.03762") → find what it builds on
3. getCitations("1706.03762") → find what builds on it
4. Recursively explore high-citation papers
5. Build a graph of the field's development

Cursor / IDE Citation Workflow

Generate proper citations while coding.

Prompt:

"Add a citation for the attention paper to my bibliography"

Agent workflow:

1. getPaperOverview(["1706.03762"]) → includes citation.bibtex and citation.apa
2. Return the BibTeX:

@article{vaswani2017attention,
  title={Attention Is All You Need},
  author={Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and ...},
  journal={arXiv preprint arXiv:1706.03762},
  year={2017}
}

Ready to paste into your .bib file — no manual formatting needed.