Find GitHub implementations of arXiv research papers β paper lookup, ranked code, search.
Copy the AI prompt to install this server into Claude Code, Cursor, or another agent β or use 1-click editor setup below.
We haven't yet run this listing's install command through our automated sandbox check. This isn't a red flag β we're steadily working through the catalog.
π‘ Paste the JSON block into your client's configuration file under mcpServers, then restart the application.
Discover GitHub implementations of research papers β from the terminal.
Search 181k+ arXiv papers and find their code on GitHub. Works for humans and AI agents alike.
Or with pipx for isolated installs:
Requires Python 3.10+. No API key needed β works anonymously at 60 requests/minute.
Use -o / --output before the subcommand:
| Format | Flag | Use case |
|---|---|---|
| Table | -o table | Human reading (default) |
| JSON | -o json | Scripts, jq pipelines, full response data |
| Quiet | -o quiet | IDs only, one per line β pipe into other commands |
| JSONL | -o jsonl | Streaming, append to files, batch processing |
| BibTeX | -o bibtex | Citation managers, LaTeX bibliographies |
| CSV | -o csv | Spreadsheets, pandas, data analysis |
The shorthand -q is equivalent to -o quiet:
You can set a default format via config so you don't need -o every time:
| Command | Description | Example |
|---|---|---|
search | Full-text paper search | codeofpaper search "vision transformers" --has-code |
paper | Paper details by arXiv ID or URL | codeofpaper paper 2010.11929 |
code | GitHub repos implementing a paper | codeofpaper code 1706.03762 |
similar | Semantically similar papers | codeofpaper similar 2010.11929 |
suggest | Autocomplete / quick lookup | codeofpaper suggest "attention" |
random | Random interesting paper | codeofpaper random --quality high |
| Command | Description | Example |
|---|---|---|
trending | Trending papers by category | codeofpaper trending --category cs.CV --sort hot |
categories | List categories or get details | codeofpaper categories cs.AI |
conferences | List all conference series | codeofpaper conferences |
conference | Papers from a specific conference | codeofpaper conference neurips_2024 --has-code |
code-drops | Recent conference papers with new code | codeofpaper code-drops --days 7 |
repo | Reverse lookup: repo β paper | codeofpaper repo google-research/vision_transformer |
open | Open paper or repo in browser | codeofpaper open 2010.11929 --code |
| Command | Description | Example |
|---|---|---|
research | Structured research overview | codeofpaper research "RL" --depth deep |
batch | Process multiple IDs from stdin/file | cat ids.txt | codeofpaper batch paper |
export | Paginated bulk export | codeofpaper export trending -o csv > out.csv |
| Command | Description | Example |
|---|---|---|
auth | Manage API key (setup/status/clear) | codeofpaper auth status |
status | Check API health and stats | codeofpaper status |
Options: --limit, --offset, --sort (relevant, recent, has_code), --has-code, --category.
Multi-step orchestrated overview of a research topic:
| Depth | Steps | API calls |
|---|---|---|
shallow | Search only + landscape statistics | 1 |
medium | Search + repos for top papers (default) | ~6 |
deep | Search + repos + similar papers from #1 | ~8 |
Process multiple queries or IDs from a file or stdin. Always outputs JSONL regardless of -o:
Supported commands: paper, search, code, similar, suggest.
Each line produces one JSON object:
Options: --delay (seconds between calls, default 0.5).
Paginated bulk export from trending, conference, or search:
Auto-paginates through results (100 per page, 0.5s delay). Options: --max (default 200), --has-code, --category, --days.
The CLI is designed for AI agent consumption. Key features:
Agents can branch on exit codes without parsing error messages:
| Exit Code | Meaning | Agent Action |
|---|---|---|
| 0 | Success | Parse stdout |
| 1 | General error | Log and report |
| 2 | Connection error | Retry with backoff |
| 3 | Not found (404) | Skip or try different ID |
| 4 | Rate limited (429) | Wait and retry |
| 5 | Auth required (401/403) | Run codeofpaper auth setup |
Responses are cached on disk for 30 minutes (~/.cache/codeofpaper/http/), so repeated calls are instant and free.
Code of Paper ships an optional Model Context Protocol server that exposes paper / code lookup as tools to any MCP-compatible agent β Claude Desktop, Cursor, Continue, Cline, Zed, etc.
Install with the mcp extra:
This adds a codeofpaper-mcp entry point that speaks MCP over stdio.
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS), %APPDATA%\Claude\claude_desktop_config.json (Windows), or ~/.config/Claude/claude_desktop_config.json (Linux):
This pulls + runs the server on demand β no global install needed. If you've already pip install'd the package, you can replace the command with just "codeofpaper-mcp" and drop args.
Most MCP clients accept the same shape:
| Tool | Purpose |
|---|---|
paper_lookup(paper_id_or_url) | Paper metadata + confident-tier repos |
code_for_paper(paper_id_or_url, limit, include_possible) | Ranked GitHub repos implementing a paper |
search_papers(query, limit, year, venue, has_code) | Free-text paper search with filters |
All tools are read-only, return plain JSON, and surface API errors as {"error": "...", "status": N} rather than throwing β agents get structured output either way.
No reviews yet β be the first to share how this listing worked for you.
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
[](https://allmcps.com/mcp/code-of-paper)<a href="https://allmcps.com/mcp/code-of-paper"><img src="https://allmcps.com/api/badge/code-of-paper?style=directory" alt="Code of Paper on AllMCPs" /></a>