RishulGupta/Doc-Ask-Engine-MCP
If you are the rightful owner of Doc-Ask-Engine-MCP and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
Doc Ask Engine MCP is a customizable, agent-powered GenAI chatbot designed for querying internal documents and data, originally developed for Naukri.com.
π¨οΈDoc Ask Engine MCP
A customizable, agent-powered GenAI chatbot for querying your own documents and data β originally built for Naukri.com to automate HR policy queries.


π What is Doc Ask Engine MCP?
Doc Ask Engine MCP is an AI-powered question-answering engine that lets you "talk" to your internal documents. It uses Generative AI and agentic reasoning, hosted on an MCP (Modular Command Processing) server, to return intelligent answers to user queries based on custom data.
Originally developed during my internship at Naukri.com, it helped answer questions like:
- π What are the office timings?
- π How many leaves can I take in a year?
- π° Whatβs the reimbursement process?
Now open-sourced and fully customizable β plug in your own PDFs to build a Q&A engine for any domain (HR, legal, medical, etc.).
β‘ Features
β
GenAI + Agentic reasoning
π PDF-based ingestion
π§ FAISS-powered similarity search
π₯οΈ API served via MCP Server
π Fully customizable with your data
π Getting Started
π§Ύ Step 1: Clone the Repository
git clone https://github.com/RishulGupta/Doc-Ask-Engine-MCP.git
cd Doc-Ask-Engine-MCP
π Step 2: Add Your PDFs
Create a folder and add your documents:
Doc-Ask-Engine-MCP/
βββ my_data/
βββ handbook.pdf
βββ code_of_conduct.pdf
βοΈ Step 3: Modify data_ingestion.py
Edit the file and change this line:
SOURCE_DIRECTORY = "my_data"
Then run:
python data_ingestion.py
This generates a FAISS index from your documents.
π§ Step 4: Start the MCP Server
python mainwithmcp.py
Youβll see the server starting up.
π¨οΈ Step 5: Ask Questions
π‘ Endpoint: http://127.0.0.2:8000/mcp/ask_faiss
Send a request with:
{
"ask_query": "What are the office working hours?"
}
You'll get a smart AI-generated response based on your uploaded documents.
π§© How It Works
- π Parse and chunk PDFs
- β¨ Embed text using GenAI
- π§ Store embeddings with FAISS
- π Retrieve relevant chunks
- π€ Agent synthesizes final answer
- π Answer served via MCP endpoint
π¨βπ» Author
Rishul Gupta
π Notes
- π Built for Naukri, generalized for everyone