Features

• PDF extraction: PyMuPDF for text layers; GLM-OCR when the PDF is scanned or image-only
• Per-document RAG: each upload gets its own Chroma collection
• Local chat: LangChain agent with inline citations; choose any installed Ollama model from the dropdown
• Markdown viewer: read extracted text, preview chunks, download markdown

Prerequisites

• Python 3.11+
• uv
• Ollama

Setup

1. Clone the repository

git clone https://github.com/dakshp26/PDFDashboardWithMCP.git
cd PDFDashboardWithMCP

2. Install dependencies

uv sync

3. Pull Ollama models

ollama pull qwen2.5:3b       # chat (or another chat model)
ollama pull nomic-embed-text # embeddings
ollama pull glm-ocr          # OCR for scanned PDFs

4. Run the app

uv run streamlit run app/main.py

Open http://localhost:8501 in your browser.

Usage

1. Upload PDF: open Upload PDF, select a file, wait for extraction to finish
2. Chat: open Chat, pick the PDF and an Ollama model, ask questions

Project Structure

app/
├── main.py                       # Entry point, page navigation
├── app_pages/
│   ├── landing.py                # Home page
│   ├── process_pdf_upload.py     # Upload + pipeline UI
│   ├── pdf_library.py            # Browse uploaded PDFs (read-only viewer)
│   └── process_pdf.py            # Viewer + chat UI
└── process_pdf/
    ├── extract.py                 # PDF → Markdown (pymupdf4llm + GLM-OCR)
    ├── pipeline.py                # Extraction pipeline with live progress
    ├── rag.py                     # Chunking, embeddings, Chroma persistence
    └── agent.py                   # LangChain agent with retriever tool
mcp_server/
└── server.py                     # MCP server (list_documents, get_document)
data/                             # Runtime data (gitignored)
├── process_pdf/                  # Saved PDFs and extracted markdown
└── process_chroma/               # Chroma vector collections (one per PDF)

[!NOTE] File-by-file breakdown, execution order, and data flow: APP_STRUCTURE.md.

Pages

Extraction progress shows in an st.status block. After processing, the Chroma collection lives in data/process_chroma/ and loads on the next run without re-extracting.

MCP Server

Two tools for MCP clients (Claude Desktop, Cursor, Claude Code):

• list_documents: indexed document collections
• get_document(document, query): semantic search over a collection

{
  "mcpServers": {
    "PDFDashboardWithMCP": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/PDFDashboardWithMCP", "mcp_server/server.py"]
    }
  }
}

{
  "mcpServers": {
    "PDFDashboardWithMCP": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/PDFDashboardWithMCP", "mcp_server/server.py"]
    }
  }
}

{
  "mcpServers": {
    "PDFDashboardWithMCP": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/PDFDashboardWithMCP", "mcp_server/server.py"]
    }
  }
}

Replace /absolute/path/to/PDFDashboardWithMCP with your clone path.

Ollama must be running with nomic-embed-text pulled before the MCP server can load collections.

Tech Stack