Provides tools for searching arXiv papers with fielded queries, retrieving full metadata and abstracts, downloading and converting papers to markdown for reading, listing category codes, and managing a local cache of downloaded papers.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@paper-mcp search for recent papers on graph neural networks"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A local, Claude-native MCP server for searching and reading
arXiv papers. Successor to the Streamlit app in paper-distill/, rebuilt as a
plain stdio MCP server with first-class Claude Code support.
No API keys, no secrets, no server-side LLM. The server only searches arXiv and converts papers to markdown — Claude (the MCP client) does all summarization and Q&A. Cheap-model summarization is handled client-side by a bundled Claude Code subagent running on Haiku.
Python 3.12+
Related MCP server: arXiv Research Assistant MCP Server
uv syncThat's it. Claude Code picks up the server automatically from the checked-in .mcp.json
when you start claude in this directory (approve the project-scoped server on first launch).
To register it elsewhere:
claude mcp add paper-mcp -- uv run --directory C:\path\to\paper-mcp python -m paper_mcpThe server is started as
python -m paper_mcprather than via thepaper-mcpconsole script: on locked-down Windows machines (AppLocker/WDAC), the generated.venv\Scripts\*.exeshims may be blocked from executing, whilepython.exeis allowed.
Tool | Description |
| Search arXiv (fielded queries |
| Full metadata + abstract for one paper (accepts IDs, |
| Download + convert the paper to markdown, paginated; also returns |
| Known arXiv category codes |
| Cache location and size |
| Delete one paper's cached files, or everything |
/mcp__paper-mcp__summarize-paper <id> [paragraphs] [audience] — audience: High School, Undergraduate, Graduate, PhD
/mcp__paper-mcp__ask-paper <id> "<question>"
/summarize-paper <arxiv-id> [paragraphs] [audience] runs the bundled
paper-summarizer subagent on Haiku, so the full
paper text never enters your main conversation context — only the summary comes back.
Env var | Default | Purpose |
|
| Where PDFs and markdown are cached |
| 2 GiB | Cache size budget (oldest files evicted first) |
|
| Log level (logs go to stderr) |
Downloads are capped at 50 MB with a 60 s timeout. Cached paper versions are immutable, so there is no TTL — the cache is size-bounded instead.
uv run python -m pytest -q # tests (no network needed)
uv run ruff check . # lint
uv run ruff format . # formatManual smoke test against the live server:
npx @modelcontextprotocol/inspector uv run python -m paper_mcpShow the location and size of the local PDF/markdown cache.
Inputs: no named parameters. Returns: cache_dir, pdf_count, total_bytes, deleted_files, +1 more
Delete cached files for one paper, or the entire cache.
Inputs: arxiv_id. Returns: cache_dir, pdf_count, total_bytes, deleted_files, +1 more
Fetch full metadata and the abstract of a single arXiv paper.
Inputs: arxiv_id (required). Returns: doi, title, abs_url, authors, +9 more
List known arXiv category codes and their display names.
Inputs: prefix. Returns: result
Download an arXiv paper, convert it to markdown, and return its text (paginated).
Also returns markdown_path: the cached file holding the FULL text, which can be Read/Grep'd directly instead of paging.
Inputs: arxiv_id (required), offset, max_chars. Returns: offset, content, arxiv_id, has_more, +4 more
Search arXiv for papers matching a query, optionally filtered by category.
Inputs: query (required), sort, categories, max_results. Returns: count, query, results
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 6 tools
Each tool has a clear, distinct purpose: searching, fetching metadata, reading full text, listing categories, and managing cache. There is no overlap or ambiguity between tools.
Tool names follow a mostly consistent verb_noun pattern with underscores (search_papers, get_paper, read_paper, clear_cache). 'cache_info' is slightly inconsistent (could be 'get_cache_info'), but the pattern is still predictable and readable.
With 6 tools, the server is well-scoped for an arXiv paper management client. It covers search, retrieval, reading, and cache management without unnecessary bloat.
The tool surface covers the core workflow: search, fetch metadata, read full text, and manage cache. A direct PDF download tool is missing, but read_paper converts to markdown, so it is not a critical gap.
Academic research MCP server for paper search, citation checks, graphs, and deep research.
Zotero MCP server for Claude and ChatGPT: search, citations, safe writes, PDF passages and pages.
Augments MCP Server - A comprehensive framework documentation provider for Claude Code
