Exa research paper search
Exa web search as standalone Agent Skills — no MCP server, just an EXA_API_KEY. Company, people, papers, code, finance, personal-site & X search + lead generation for Claude Code, Codex, Cursor, OpenCode, Gemini CLI, and any SKILL.md agent.
npx -y skills add CodeAlive-AI/exa-skills --skill exa-research-paper-searchAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Find academic papers with Exa — by topic, author, recency, or as surveys/reviews. Use for literature reviews, finding seminal or state-of-the-art papers, tracking recent preprints (arXiv), or building a reading list. Runs a local script against the Exa API; no MCP server required.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
2.9 KB, as published. Nobody here has run it
Research Paper Search (Exa)
Find academic papers and preprints via Exa's research-paper index (arXiv and beyond). Calls the Exa REST API through a local script; no MCP server needed, only an EXA_API_KEY.
Setup (once)
export EXA_API_KEY=your-key # or EXA_API_KEY=... in a .env at the repo root
Get a key at https://dashboard.exa.ai/api-keys. Shared details: exa-native-base.
Run it
python scripts/research_paper_search.py "<topic / method>" [-n N] [--json]
Examples:
python scripts/research_paper_search.py "sparse attention mechanisms for long context transformers" -n 12
python scripts/research_paper_search.py "diffusion models comprehensive survey review" -n 10
python scripts/research_paper_search.py "Geoffrey Hinton forward-forward algorithm" -n 5
Defaults to category=research paper.
Query patterns
# By topic
python scripts/research_paper_search.py "category:research paper retrieval augmented generation robustness" -n 12
# Survey / review papers (best entry point into a field)
python scripts/research_paper_search.py "category:research paper <topic> comprehensive survey review" -n 10
# By author
python scripts/research_paper_search.py "category:research paper <author name> <topic>" -n 5
# By recency — encode the year(s) in the query (don't rely on date filters alone)
python scripts/research_paper_search.py "category:research paper large language model agents 2025 2026" -n 15
To find seminal papers: search for survey/review papers first, then deep-read them to extract the foundational references they cite.
Token isolation (for a literature review)
Decompose into sub-topics and dispatch a subagent per branch (e.g. architectures, training, evaluation, limitations). Each runs 3-5 paper searches and returns a compact table (title · authors · year · venue/arXiv id · one-line contribution). Merge, dedupe by title/arXiv id, then synthesize.
Override categories with -c
research paper (default) · pdf (when the paper is a standalone PDF) · news (coverage of a result). For code accompanying a paper, use exa-code-context.
After you get results
- Deep-read abstracts/sections:
python ../exa-native-base/scripts/exa.py contents <arxiv-url> --text. - For a narrative review, see
references/synthesis.mdinexa-native-base— organize by theme, cite every claim, surface disagreements. - Deliver a table (title · authors · year · link · contribution) or a themed synthesis with citations.