name: search_skill description: Academic literature search skill for finding papers from arXiv, Semantic Scholar, and HuggingFace version: 1.0.0 author: PaperAgent Team

Academic Literature Search Skill

This skill enables you to search for academic papers and literature from multiple sources. Use this skill when you need to find relevant research papers for a topic.

Available Tools

You have access to the following search tools (registered in the Toolkit):

1. search_arxiv

Search for papers on arXiv preprint server.

Parameters:

• query (str, required): The search query string
• max_results (int, optional): Maximum number of results (default: 10, max: 50)
• sort_by (str, optional): Sort by "relevance" or "date"

Returns: List of papers with title, authors, abstract, URL, year, and categories.

Example:

search_arxiv(query="large language models", max_results=10, sort_by="relevance")

2. search_semantic_scholar

Search for papers on Semantic Scholar with citation information.

Parameters:

• query (str, required): The search query string
• max_results (int, optional): Maximum number of results (default: 10)
• year_from (int, optional): Filter papers from this year
• year_to (int, optional): Filter papers to this year

Returns: List of papers with title, authors, abstract, URL, year, citation count, and venue.

Example:

search_semantic_scholar(query="transformer architecture", max_results=20, year_from=2020)

3. search_huggingface_papers

Search for papers on HuggingFace Papers.

Parameters:

• query (str, required): The search query string
• max_results (int, optional): Maximum number of results (default: 10)

Returns: List of papers with title, authors, abstract, and URL.

Example:

search_huggingface_papers(query="multimodal learning", max_results=15)

4. academic_search

Multi-source academic search that combines results from multiple sources.

Parameters:

• query (str, required): The search query string
• sources (list[str], optional): Sources to search (default: ["arxiv", "semantic_scholar"])
• max_results_per_source (int, optional): Max results per source (default: 5)

Returns: Combined and deduplicated list of papers from all sources.

Example:

academic_search(query="neural networks", sources=["arxiv", "semantic_scholar"], max_results_per_source=10)

Search Strategies

Topic-based Search

For exploring a research topic:

1. Start with broad query terms
2. Use academic_search to get papers from multiple sources
3. Identify key papers and refine search based on terminology found

Author-based Search

For tracking specific researchers:

1. Include author name in query: "Author Name" topic
2. Use Semantic Scholar for better author disambiguation

Citation-based Search

For finding related work:

1. Search for seed papers on the topic
2. Use citation tools to trace references and citing papers

Time-based Search

For finding recent advances:

1. Use year_from parameter in Semantic Scholar
2. Use sort_by="date" in arXiv

Best Practices

1. Formulate precise queries: Use specific technical terms
2. Combine sources: Different databases have different coverage
3. Filter by year: For up-to-date research, filter recent papers
4. Check citation counts: High citation papers are often foundational
5. Read abstracts first: Quickly assess relevance before deep reading
6. Save search results: Keep track of found papers for later reference

Output Format

Search results are returned in markdown format with:

• Paper title
• Authors (first 3 + "et al." if more)
• Publication year
• Citation count (when available)
• Venue (when available)
• Abstract (truncated to 500 chars)
• URL/DOI link