Skip to content
MCP server

Trove

Search papers, read open full text and trace citations across open scholarly APIs

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted
Runs
Locally
Version
0.1.11
Listing updated
today

Similar MCP servers

About Trove

Trove is an open-source MCP server for health and science. It runs locally on your machine and is started by your AI client.

Trove draws on OpenAlex for metadata and citations, Semantic Scholar for similarity, arXiv for preprints and full text, Unpaywall for open-access lookup by DOI, PubMed, CORE as a full-text fallback and the Hugging Face papers API. The author calls it deliberately conservative: when a provider cannot support an answer, tools return partial, error or unavailable with warnings rather than a guess.

Some tools have prerequisites. Trending papers by citation velocity need at least two local snapshot dates, so a fresh install usually returns unavailable until a scheduled sync has built history. Similar-paper recommendations use Semantic Scholar only, with no keyword fallback, and need a key to be reliable. Reference lists can be missing for some large arXiv preprints, in which case the full-text tool is suggested. Setup-sensitive tools can be hidden with an environment variable.

It runs over stdio with npx, or as a Streamable HTTP server protected by a Bearer token, caches results in a local SQLite file and returns a structured envelope with status, warnings and provenance. No API keys are required to start; contact emails improve polite access to OpenAlex and Unpaywall.

Install

Run in a terminal

claude mcp add trove-mcp --env SEMANTIC_SCHOLAR_API_KEY='<SEMANTIC_SCHOLAR_API_KEY>' --env CORE_API_KEY='<CORE_API_KEY>' -- npx -y trove-mcp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "trove-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "trove-mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "<SEMANTIC_SCHOLAR_API_KEY>",
        "CORE_API_KEY": "<CORE_API_KEY>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "trove-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "trove-mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "<SEMANTIC_SCHOLAR_API_KEY>",
        "CORE_API_KEY": "<CORE_API_KEY>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "trove-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "trove-mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "<SEMANTIC_SCHOLAR_API_KEY>",
        "CORE_API_KEY": "<CORE_API_KEY>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "trove-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "trove-mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "<SEMANTIC_SCHOLAR_API_KEY>",
        "CORE_API_KEY": "<CORE_API_KEY>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "trove-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "trove-mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "<SEMANTIC_SCHOLAR_API_KEY>",
        "CORE_API_KEY": "<CORE_API_KEY>"
      }
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Configuration

  • TROVE_CONTACT_EMAILContact email for polite OpenAlex usage
  • UNPAYWALL_EMAILEmail required to enable Unpaywall open-access lookup
  • TROVE_DB_PATHPath to the local SQLite cache and citation-snapshot database
  • SEMANTIC_SCHOLAR_API_KEYNeeded for reliable find_similar_papers
  • CORE_API_KEYImproves CORE full-text throughput

Tools this server provides (10)

Taken from the project documentation; the running server may differ.

search_papers
Multi-source search with deduplication and ranking
get_trending_papers
Ranks topic papers by citation velocity
get_paper
Resolves a paper by DOI, arXiv, S2, OpenAlex, PubMed or title
get_full_text
Fetches open full text with fallbacks
get_citations
Lists papers that cite a target
get_references
Lists papers a target references
find_similar_papers
Semantic Scholar recommendations
get_author
Author profile and impact metrics
trace_idea
Traces the origin of an idea
build_literature_map
Builds a literature map

What it does

Multi-source search
Searches OpenAlex, Semantic Scholar, arXiv, PubMed and CORE with deduplication.
Open full text
Falls back from arXiv to OpenAlex PDFs, Unpaywall and CORE with chunked output.
Citations and references
Lists papers that cite a target and papers it references.
Fail-closed answers
Returns unavailable or partial results instead of approximations.
Citation snapshots
Builds local history to rank trending papers by citation velocity.
Local cache
Stores results and snapshots in SQLite.
Provenance envelope
Each result carries status, warnings and provenance.

Good for

  • Find papers on a topic and compare competing methods
  • Trace the origin of an idea through citations
  • Read the open full text of a preprint
  • Build a literature map for a research area
  • Look up an author's profile and impact

Pricing

Open source

Trove FAQ

Do I need API keys?

None are required to start; optional keys improve similar-paper and CORE results.

Why does trending return unavailable at first?

It needs at least two local citation snapshot dates, so run the sync command on a schedule.

Can I hide tools that need extra setup?

Yes, with TROVE_DISABLED_TOOLS.

Can it run over HTTP?

Yes, in streamable HTTP mode with a Bearer token.

Is Trove free?

Yes. Trove is open source, so you can use and self-host it for free.

How do I add Trove to Claude Code?

Run this command in your terminal: claude mcp add trove-mcp --env SEMANTIC_SCHOLAR_API_KEY='<SEMANTIC_SCHOLAR_API_KEY>' --env CORE_API_KEY='<CORE_API_KEY>' -- npx -y trove-mcp

What are the best alternatives to Trove?

The closest alternatives to Trove are Tooluniverse, Oya Browser, BasalFit, Clinicaltables and Complex Portal.

Where this comes from

  • Description: written from the project's README.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers