Scholar Feed MCP Server

Access to all AI/ML papers on ArXiv (600k+) with semantic search + more

Documentation

Scholar Feed MCP Server

Search 600,000+ CS/AI/ML research papers with LLM-powered novelty analysis from Claude Code, Cursor, or any MCP client.

Scholar Feed indexes arXiv papers daily and ranks them using a multi-signal scoring system (recency, citation velocity, institutional reputation, code availability). Each paper has an LLM-generated summary and novelty score.

Quick Start

npx scholar-feed-mcp init

This interactive wizard will:

Optionally ask for an API key (or skip for anonymous access)
Detect your MCP client (Claude Code, Cursor, or Claude Desktop)
Write the config and verify the connection

No API key required. Anonymous access gives you 100 calls/day — enough for a typical research session. For higher limits (1,000/day per account), get a free key at scholarfeed.org/settings.

Try asking: "Search for recent papers on test-time compute scaling"

Migrating from v1.x to v3.0.0

v3.0.0 is a hard cutover — there is no deprecation window. The previous v1.x tool surface was consolidated down to 8 focused tools. Existing agents using removed tools will need to update their calls before upgrading.

Breaking changes — full replacement table

Removed tool (v1.x)	v3 replacement
`find_similar(arxiv_id=id)`	`search_papers(anchor_paper_id=id)`
`find_citations_about(arxiv_id=X, query=Q)`	`search_papers(scope_to_citations_of=X, q=Q)`
`whats_trending(category=C)`	`search_papers(sort='trending', category=C)`
`batch_lookup(arxiv_ids=[...])`	`get_paper(arxiv_ids=[...])`
`export_bibtex(arxiv_ids=[...])`	`get_paper(arxiv_ids=[...], format='bibtex')`
`discover_authors(q=Q)`	`find_author(q=Q)`
`get_author(author_id=id)`	`find_author(id=id)`
`compare_methods(models=[...])`	Use the `/compare-methods` skill (see scholarfeed.org/skills)
`field_guide(topic=T)`	`get_field_orientation(topic=T)` for cheap retrieval; full orientation via the `/field-guide` skill
`check_connection`	Removed — errors signal connectivity. No action needed; remove any health-check calls.
`fetch_repo(arxiv_id=id)`	Removed — 0 observed calls in production. Backend route preserved for skill use.

The 9 v3 tools

Tool	Description
`search_papers`	Semantic + keyword search. Now also handles similar-paper discovery (`anchor_paper_id`), citation-scoped search (`scope_to_citations_of`), and trending (`sort='trending'`).
`get_paper`	Full paper details by arXiv ID. Now also handles batch lookup (`arxiv_ids=[...]`) and BibTeX export (`format='bibtex'`).
`get_citations`	Citation graph — outgoing refs or incoming citations (`direction='citing'/'cited'`).
`fetch_fulltext`	Extract results/experiments sections from LaTeX source.
`find_author`	Find authors by name/topic query (`q=`) or retrieve a profile by ID (`id=`). Merges the former `discover_authors` + `get_author`.
`co_author_graph`	Co-authorship neighborhood for an author — edges derived live from the citation graph.
`embed_text`	Get a 768-dim Gemini embedding for a text string (useful for HyDE and custom similarity).
`get_field_orientation`	Cheap retrieval orientation for a research area — top papers, subfields, open problems. No Pro quota.
`get_foundational_lineage`	The foundational work for a paper's niche via the citation graph — niche_roots → field_level → discipline (collapsed landmarks), with `cited_by_in_niche` evidence. Surfaces canonical anchors semantic search misses. No Pro quota.

New in v3.1.0:

get_foundational_lineage — paper-anchored citation-graph lineage (consensus-then-lift). Answers the relative question: what is foundational for THIS paper's specific niche, not the obvious global landmarks. Complements get_field_orientation (topic-anchored, retrieval-only).

New in v3.0.0 (reaching npm for the first time):

find_author — merged discover_authors + get_author into a single tool with exactly-one-of (q or id) semantics.
get_field_orientation — the cheap-retrieval half of the former field_guide, demoted from tool to skill in v3 but re-added as a lightweight tool (0.6 × citation score + 0.4 × cosine similarity; no DeepSeek, no Pro quota). Pairs with the /field-guide skill for deeper orientation.
co_author_graph and embed_text — shipped in the local v2.1.0 build but never published to npm; v3.0.0 is the first public release of both.

What You Can Do

Technology scouting — "What novel research on retrieval-augmented generation was published this month?"

Literature review — "Find papers similar to 2401.04088 and export their BibTeX"

Trend monitoring — "What's trending in cs.CV this week? Summarize the top 3."

Author discovery — "Who are the top researchers working on efficient LLM inference?"

Field orientation — "Give me an orientation report on sparse mixture-of-experts architectures."

Manual Installation

The fastest path is npx scholar-feed-mcp init, which auto-detects your client and writes the config. To do it by hand, find your client below.

Every client launches the same stdio server — npx -y scholar-feed-mcp. Most use the identical mcpServers JSON block; only the config-file location and the wrapper key change. The standard block (used by Cursor, Claude Desktop, Windsurf, Cline/Roo, Gemini CLI, LM Studio, and JetBrains) is:

{
  "mcpServers": {
    "scholar-feed": {
      "command": "npx",
      "args": ["-y", "scholar-feed-mcp"]
    }
  }
}

API key (optional, for 1,000 calls/day): add "env": { "SF_API_KEY": "sf_your_key_here" } to the server entry. Get a free key at scholarfeed.org/settings.

Claude Code

# Without API key (anonymous, 100 calls/day)
claude mcp add scholar-feed -- npx -y scholar-feed-mcp

# With API key (1,000 calls/day per account)
claude mcp add scholar-feed -e SF_API_KEY=sf_your_key_here -- npx -y scholar-feed-mcp

Cursor (`.cursor/mcp.json`)

Use the standard block above in .cursor/mcp.json (project) or ~/.cursor/mcp.json (global), then restart Cursor.

Claude Desktop (`claude_desktop_config.json`)

Settings → Developer → Edit Config, paste the standard block, then restart. File location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS), %APPDATA%\Claude\claude_desktop_config.json (Windows).

VS Code — GitHub Copilot (`.vscode/mcp.json`)

VS Code uses a servers key (not mcpServers) and an explicit type. Requires Copilot agent mode. You can also run MCP: Add Server from the Command Palette.

{
  "servers": {
    "scholar-feed": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "scholar-feed-mcp"]
    }
  }
}

Windsurf (`~/.codeium/windsurf/mcp_config.json`)

In Cascade, click the MCP icon → Configure (opens ~/.codeium/windsurf/mcp_config.json), paste the standard block, then click refresh.

Cline / Roo Code

Click the MCP Servers icon in the sidebar → Configure / Edit, then paste the standard block into cline_mcp_settings.json. Cline and Roo Code share this format.

Zed (`settings.json`)

Zed uses a context_servers key, and the "source": "custom" line is required — without it, Zed silently skips the entry.

{
  "context_servers": {
    "scholar-feed": {
      "source": "custom",
      "command": "npx",
      "args": ["-y", "scholar-feed-mcp"]
    }
  }
}

Continue (`~/.continue/config.yaml`)

Continue uses YAML, with mcpServers as a list. Add to ~/.continue/config.yaml (global) or .continue/config.yaml (workspace).

mcpServers:
  - name: scholar-feed
    type: stdio
    command: npx
    args:
      - "-y"
      - scholar-feed-mcp

Gemini CLI (`~/.gemini/settings.json`)

Add the standard block to ~/.gemini/settings.json (or project .gemini/settings.json).

JetBrains — PyCharm / IntelliJ (AI Assistant)

Settings → Tools → AI Assistant → Model Context Protocol (MCP) → Add → As JSON, paste the standard block, then apply. Requires AI Assistant 2025.1+.

LM Studio (`~/.lmstudio/mcp.json`)

Program tab → Install → Edit mcp.json (~/.lmstudio/mcp.json), paste the standard block, then save. LM Studio follows Cursor's mcp.json notation.

Project-scoped (`.mcp.json`)

{
  "mcpServers": {
    "scholar-feed": {
      "command": "npx",
      "args": ["-y", "scholar-feed-mcp"],
      "env": { "SF_API_KEY": "${SF_API_KEY}" }
    }
  }
}

Windows note: for any of the JSON configs above, use "command": "cmd" and "args": ["/c", "npx", "-y", "scholar-feed-mcp"].

Any other MCP client

Scholar Feed is a standard stdio MCP server, so any MCP-compatible client works — use the standard mcpServers block above with command: npx, args: ["-y", "scholar-feed-mcp"].

Available Tools (22)

Core Search & Discovery

Tool	Description	Key Parameters
`search_papers`	Semantic + keyword search with filters. Also does similar-paper discovery, citation-scoped search, and trending.	`q`, `category`, `novelty_min`, `days`, `sort`, `anchor_paper_id`, `scope_to_citations_of`, `mode`, `method_category`, `task`, `dataset`, `contribution_type`, `task_category`, `cursor`, `limit`
`get_paper`	Get full paper details by arXiv ID. Also handles batch lookup and BibTeX export.	`arxiv_ids`, `format`, `fields`, `verbose`
`get_citations`	Citation graph (outgoing refs or incoming citations)	`arxiv_id`, `direction`, `limit`, `fields`
`fetch_fulltext`	Extract results/experiments from LaTeX source	`arxiv_id`

Authors

Tool	Description	Key Parameters
`find_author`	Find researchers by topic/name query, or retrieve a profile by ID.	`q`, `id`, `field`, `limit`
`co_author_graph`	Co-authorship neighborhood for an author	`author_ids`, `window_years`

Embeddings

Tool	Description	Key Parameters
`embed_text`	Get a 768-dim Gemini embedding for text (for HyDE and custom similarity)	`text`, `task_type`

Research

Tool	Description	Key Parameters
`get_field_orientation`	Cheap retrieval orientation for a research area — top papers, subfields, open problems. No Pro quota.	`topic`, `limit`
`get_foundational_lineage`	Foundational work for a paper's niche via the citation graph (consensus-then-lift): niche_roots → field_level → discipline, with `cited_by_in_niche` evidence. Surfaces canonical anchors semantic search misses. No Pro quota.	`anchor_paper_id`, `scope`, `generality_ceiling`, `limit`

Library, Collections, Watches & Gap Analysis (require `SF_API_KEY`)

These MUTATE or read the authenticated user's account. The core read/search tools above work anonymously; these need a key.

Tool	Description	Key Parameters
`save_paper`	Bookmark a paper to your library (idempotent; feeds personalization).	`arxiv_id`
`unsave_paper`	Remove a paper from your library (idempotent).	`arxiv_id`
`like_paper`	"More like this" calibration signal for the For You feed (insert-only).	`arxiv_id`
`list_library`	List your saved papers, newest first.	`limit`, `page`
`list_collections`	List collections with paper counts.	—
`create_collection`	Create a named collection (get-or-create; no error on duplicate).	`name`
`add_to_collection`	Add a paper to a collection by name or id (also auto-saves).	`arxiv_id`, `collection_name`, `collection_id`
`remove_from_collection`	Remove a paper from a collection (stays saved).	`arxiv_id`, `collection_name`, `collection_id`
`create_watch`	Standing daily-evaluated saved search; get-or-create by name. Exactly one seed selector.	`name`, `novelty_min`, `q`, `collection_name`, `collection_id`, `anchor_paper_id`, `scope_to_citations_of`, `author_id`, `category`
`list_watches`	List watches with summary, `last_evaluated_at`, and `pending_hits`.	—
`check_watches`	Pull new matches since the last digest (read-only, idempotent).	`watch_name`, `watch_id`, `limit`
`delete_watch`	Delete a watch by name or id (idempotent).	`name`, `watch_id`
`find_gaps`	"What am I missing?" for a collection or topic — foundational + frontier work you haven't saved (read-only, Pro).	`collection_name`, `collection_id`, `topic`, `scope`, `limit`

Novelty Score

Every paper has an llm_novelty_score from 0.0 to 1.0:

Range	Meaning	Example
0.7+	Paradigm shift or broad SOTA	New architecture that changes the field
0.5-0.7	Novel method with strong results	New training technique with clear gains
0.3-0.5	Incremental improvement	Applying known method to new domain
<0.3	Survey, dataset, or minor extension	Literature review, benchmark release

Use novelty_min: 0.5 in search_papers to filter for genuinely novel work.

Rate Limits

Endpoint	Limit
`search_papers`	30/min
`get_paper`	30/min
`get_citations`	30/min
`fetch_fulltext`	10/min
`find_author`	20/min
`co_author_graph`	20/min
`embed_text`	30/min
`get_field_orientation`	20/min
`get_foundational_lineage`	20/min
`find_gaps`	20/min

Responses include X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset headers.

Example Response

search_papers with q: "attention mechanism" returns:

{
  "papers": [
    {
      "arxiv_id": "2401.04088",
      "title": "Attention Is All You Need (But Not All You Get)",
      "authors": ["A. Researcher", "B. Scientist"],
      "year": 2024,
      "categories": ["cs.LG", "cs.AI"],
      "primary_category": "cs.LG",
      "arxiv_url": "https://arxiv.org/abs/2401.04088",
      "has_code": true,
      "github_url": "https://github.com/example/repo",
      "citation_count": 42,
      "rank_score": 0.73,
      "llm_summary": "Proposes a sparse attention variant that reduces compute by 60% while matching dense attention accuracy on 5 benchmarks.",
      "llm_novelty_score": 0.55
    }
  ],
  "total": 1847,
  "page": 1,
  "limit": 20,
  "next_cursor": "eyJzIjogMC43MywgImlkIjogIjI0MDEuMDQwODgifQ=="
}

Pass next_cursor back to get the next page (keyset pagination — more stable than page numbers for large result sets).

Environment Variables

Variable	Required	Default	Description
`SF_API_KEY`	No	—	Your Scholar Feed API key (starts with `sf_`). Without it, runs in anonymous mode (100 calls/day).
`SF_API_BASE_URL`	No	Production URL	Override API base URL

Development

npm install
npm run build      # Build to build/
npm run dev        # Watch mode
npm run typecheck  # Type check without emitting
npm test           # Run tests

Contributing

See CONTRIBUTING.md for guidelines.

Troubleshooting

"Authentication failed: your SF_API_KEY is invalid" The key may have been revoked. Generate a new one at scholarfeed.org/settings. Or remove the key to use anonymous mode.

"Rate limit exceeded" or "Anonymous daily limit exceeded" Anonymous mode allows 100 calls/day. Get a free API key at scholarfeed.org/settings for 1,000 calls/day per account.

Tool calls time out or fail silently Ensure Node.js 18+ is installed (node --version). Older versions lack the native fetch API.

Stale npx cache If you're stuck on an old version after an update: npx --yes scholar-feed-mcp@latest

Windows: "command not found" Use "command": "cmd" with "args": ["/c", "npx", "-y", "scholar-feed-mcp"] in your MCP config.

License

MIT

Scholar Feed MCP Server

Documentation