Hosted, nothing to install
Point your client at one URL. No local server, no database to keep up to date.
Feature · Paperlist MCP
Connect the hosted Paperlist MCP server to Claude Code, Cursor, Codex, ChatGPT, or any MCP client. Your AI can then search 210,000+ papers from 30+ top conferences and pull up the details it needs.
Free to browse · No account needed
Paperlist features
Paperlist MCP is a hosted MCP server. Add its URL to your AI client and the AI can search papers from 30+ top conferences, filter by conference, year, and decision type, and fetch abstracts, PDFs, and official links.
Key points
Point your client at one URL. No local server, no database to keep up to date.
Results come from conference paper lists, so every paper has a venue and year to cite.
Claude Code, Cursor, Codex, ChatGPT, and other MCP clients.
The server exposes the same data you see on Paperlist.
General web search mixes preprints, blog posts, and duplicates. Paperlist MCP answers from conference papers only, with structured fields your AI can filter on instead of guessing from page text.
Paperlist MCP
Paperlist runs a hosted MCP server. Connect it to Claude Code, Cursor, Codex, ChatGPT, or any MCP client, and your AI can look up papers from 30+ top conferences while it works.
What your AI can do
Server URL
https://paperlist.ai/mcpSign in with your free Paperlist account when your client asks, or create an API key in your dashboard settings.
Connect your client
Claude Code
Run in your terminal, then run /mcp in Claude Code to sign in. Or pass an API key with --header
claude mcp add --transport http paperlist https://paperlist.ai/mcp
# or, with an API key
claude mcp add --transport http paperlist https://paperlist.ai/mcp \
--header "Authorization: Bearer pl_your_key"Cursor
Add to .cursor/mcp.json. Cursor opens Paperlist sign-in. To use an API key, add headers
{
"mcpServers": {
"paperlist": {
"url": "https://paperlist.ai/mcp"
}
}
}
// or, with an API key
{
"mcpServers": {
"paperlist": {
"url": "https://paperlist.ai/mcp",
"headers": { "Authorization": "Bearer pl_your_key" }
}
}
}Codex
Add to ~/.codex/config.toml, then run codex mcp login paperlist. Or set bearer_token_env_var and export your API key
[mcp_servers.paperlist]
url = "https://paperlist.ai/mcp"
# optional: use an API key instead of codex mcp login
# bearer_token_env_var = "PAPERLIST_API_KEY"ChatGPT
Add as a custom connector with this URL and choose OAuth. ChatGPT opens Paperlist sign-in
https://paperlist.ai/mcpTools
list_conferencesConference slugs, full names, domains and years. Can also return one edition's topics.
Try asking: Which years of CoRL does Paperlist have?
topics_for { conference, year }search_papersFull-text search with filters for conference, year, decision, topic and author.
Try asking: Find Oral papers on video diffusion at ICLR 2025 and NeurIPS 2024.
query stringconferences string[]year_from numberyear_to numberdecision string[]topic stringauthor stringsort relevance | citations | newestlimit numbercursor stringget_paperOne paper's full record: abstract, authors, venue, decision, citations and links.
Try asking: Open the top result and summarize its method.
id string · requiredlist_papersBrowse one conference edition, optionally by decision or topic.
Try asking: List the Spotlight papers at ICML 2025.
conference string · requiredyear number · requireddecision string[]topic stringlimit numbercursor stringlist_topicsAll 207 research topics with trend stats, grouped by area, with paper counts.
Try asking: Which reinforcement learning topics have the most papers since 2023?
get_topic_statsHow a research topic grew at NeurIPS/ICLR/ICML/CVPR 2023–2026: counts by year and venue, growth, sub-themes, representative papers.
Try asking: Use paperlist to show how world models papers grew at NeurIPS 2023–2026.
topic string · requiredfind_similar_papersPapers related to one you already have, to grow a related-work list.
Try asking: Find papers similar to this one from 2024 onward.
id string · requiredconferences string[]year_from numberyear_to numberlimit numbercursor stringFor clients that expect these two names, such as ChatGPT deep research.
searchSimple keyword search that returns ids for fetch.
Try asking: Search Paperlist for retrieval-augmented generation benchmarks.
query string · requiredfetchRead one paper by an id from search.
Try asking: Fetch the second result and give me its citation.
id string · requiredSource data
Every paper returned links to its PDF or official page, so you can check what the AI cites.
More ways to browse papers from top AI conferences.
Connect
Use the server URL above in Claude Code, Cursor, Codex, ChatGPT, or another MCP client.