Neura Market
connerlambden/bgpt-mcp logo

connerlambden/bgpt-mcp

Free

搜索科学论文，获取从全文研究中提取的结构化实验数据，每篇论文返回25+字段，包括方法、结果、样本量、局限性和质量评分。

FreeFree tier
Inputs: text
Type
Open Source
Company
connerlambden
Visit Website GitHub (40 stars)

About connerlambden/bgpt-mcp

BGPT is a remote Model Context Protocol (MCP) server and traditional JSON/HTTP API that provides AI assistants and Python applications with structured data from scientific papers. Unlike typical search tools that return only titles and abstracts, BGPT extracts raw experimental data from full-text studies, delivering 25+ metadata fields per paper including methods, results, conclusions, quality scores, sample sizes, limitations, and more. It supports MCP transports (SSE and Streamable HTTP), a REST API, and direct Python integration. A free tier offers 50 results without an API key. The tool also includes evidence dashboards (Plotly), claim-interrogation prompts, and a prompt gallery for scientific RAG and literature review agents.

Key Features

Returns 25+ structured metadata fields per paper (methods, results, sample sizes, limitations, quality scores)
Extracts raw experimental data from full-text studies
Supports MCP (SSE and Streamable HTTP) and REST API
Can be used from any MCP-compatible client (Claude, Cursor, etc.) or plain Python
Generates Plotly evidence dashboards for study methods, samples, limitations, and falsifiability
Includes a prompt gallery for scientific RAG, literature review agents, and claim interrogation
Free tier provides 50 search results without requiring an API key

Pros & Cons

Pros
  • Returns structured data beyond titles and abstracts, enabling deeper analysis
  • Multiple integration options: MCP, REST, and Python
  • Free tier available with no API key required
  • Designed to help AI agents evaluate evidence quality and falsifiability
  • Open-source with clear documentation and examples
Cons
  • Free tier limited to 50 search results
  • Requires a compatible MCP client or API integration for best use
  • Database coverage and update frequency not explicitly stated

Best For

Search scientific papers by query (e.g., clinical trials, gene editing methods)Retrieve detailed experimental metadata for evidence-based reasoningInterrogate scientific claims by examining study limitations, sample sizes, and quality scoresGenerate literature review summaries and evidence dashboardsIntegrate structured paper data into AI agent workflows (RAG, decision support)

FAQ

Do I need an API key to use BGPT?
No, the free tier works without an API key and provides 50 search results. An API key may be required for higher usage limits.
What integration options does BGPT support?
BGPT supports remote MCP connections via SSE and Streamable HTTP, a traditional REST API, and direct Python scripting. It can be used with clients like Claude Desktop, Cursor, Claude Code, and others.
How is BGPT different from other paper search tools?
BGPT extracts structured experimental data from full-text studies, returning 25+ metadata fields per paper (methods, results, sample sizes, limitations, quality scores) rather than just titles and abstracts.