RAG-based MCP Server
English
This package provides a Retrieval Augmented Generation (RAG)-based Model Context Protocol (MCP) server, which generates contextually accurate responses by leveraging PDF data. It uses OpenAI's embedding model (text-embedding-3-small) to create embeddings from PDF documents and performs hybrid search (keyword + semantic search) to retrieve relevant content.
Installation
-
Install UV (if you haven't already):
-
For macOS or Linux:
curl -LsSf https://astral.sh/uv/install.sh | sh
-
For Windows (using PowerShell):
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
-
Configuration
To use this server in the Claude app, add the following configuration to your claude_desktop_config.json file:
{
"mcpServers": {
"rag-mcp": {
"command": "uvx",
"args": [
"mcp-hybrid-search",
"path/to/your/allowed/folder"
],
"env": {
"OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
}
}
}
}
- Replace
"path/to/your/allowed/folder"with the actual folder path where your PDF files will be stored. - Create a subfolder named
datawithin the specified folder, and place all PDF documents you want to use into thisdatafolder.
Usage
- Complete the installation and configuration steps described above.
- Launch or restart the MCP host application (e.g., the Claude app). The RAG server will automatically start and process PDF embeddings.
- The server will now generate contextually accurate and relevant responses by referencing the PDF documents stored in the
datafolder.
License
MIT
한국어
이 패키지는 PDF 데이터를 활용하여 보다 정확한 답변을 생성할 수 있는 RAG(Retrieval Augmented Generation) 기반 MCP(Model Context Protocol) 서버입니다. OpenAI 임베딩 모델(text-embedding-3-small)을 사용해 PDF 데이터를 임베딩한 후, 하이브리드 서치 (키워드 + 시멘틱 서치) 방식으로 검색 기능을 제공합니다.
설치 방법
-
UV 설치하기 (아직 설치하지 않았다면)
-
macOS 또는 Linux의 경우:
curl -LsSf https://astral.sh/uv/install.sh | sh
-
Windows의 경우 (PowerShell에서 실행):
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
-
설정 방법
Claude 앱에서 이 서버를 사용하려면, claude_desktop_config.json 파일에 다음 설정을 추가해야 합니다:
{
"mcpServers": {
"rag-mcp": {
"command": "uvx",
"args": [
"mcp-hybrid-search",
"접근을 허용할 폴더 경로"
],
"env": {
"OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
}
}
}
}
"접근을 허용할 폴더 경로"부분에는 실제로 PDF 파일을 보관할 경로를 적어주세요.- 해당 경로 아래에
data라는 이름의 폴더를 만들어 주시고, 이 폴더 안에 자료로 사용할 PDF 파일을 넣어주세요.
사용 방법
- 위의 설치와 설정 과정을 모두 완료한 후,
- MCP 호스트 애플리케이션(예 : Claude 앱)을 실행하거나 재시작하면 PDF 데이터 임베딩 후 자동으로 RAG 서버가 활성화됩니다.
- 이제 서버가
data폴더에 넣어둔 PDF 파일의 내용을 참고하여 보다 정확하고 맥락에 맞는 답변을 생성해줍니다.
라이센스
MIT
Metadata
Release files for mcp-hybrid-search 1.2.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| mcp_hybrid_search-1.2.2.tar.gz | 7.9 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| mcp_hybrid_search-1.2.2-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 17.2 kB
Release files / mcp_hybrid_search-1.2.2.tar.gz
| Download URL | mcp_hybrid_search-1.2.2.tar.gz |
|---|---|
| Size | 7.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3be2bfe93920668834c24bf93544486b5d7b51fbf594ca5f82c8b9977717cb3c
|
|
BLAKE2b-256 checksum How to use checksums |
329ecbb7c8cb4fc497ffbc3f99d4abbf6ef30c0ea186f1828068fadcb09e6b81
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.1
|
Release files / mcp_hybrid_search-1.2.2-py3-none-any.whl
| Download URL | mcp_hybrid_search-1.2.2-py3-none-any.whl |
|---|---|
| Size | 9.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
0aa3a0df0609d26a438ebb55e63d6028b211858e55be8fef716dacc644b4979a
|
|
BLAKE2b-256 checksum How to use checksums |
fde8491bcdf76e6bec96fa18010d095456fb83bff2638af1eb24bcdfe7fdaa47
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.1
|