Contract Comparisons

MCP server with 2 agents to compare contracts to QIPO

python

GitHub Stars

python

Language

7 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ec49ca-nlp-project-contract-comparison": {
      "command": "bash",
      "args": [
        "-lc",
        "source venv/bin/activate && python3 -m uvicorn backend.server.mcp_server:app --host 0.0.0.0 --port 8000"
      ],
      "env": {
        "ENV": "development",
        "PORT": "8000",
        "LOG_LEVEL": "INFO",
        "ALLOWED_ORIGINS": "*"
      }
    }
  }
}

You set up a multi-agent MCP Server that orchestrates several language models to process complex queries against uploaded PDFs and external data sources. It provides a fast API backbone, a modern web UI, automatic agent discovery, and flexible provider support, enabling you to query contracts and related documents with intelligent routing and synthesis.

How to use

Start the MCP Server and its frontend, then open the chat interface to interact with the system. You can switch between provider and model options, upload PDFs, and ask questions that are answered by combining results from multiple agents.

How to install

Prerequisites ensure you have the right runtimes and tooling to run both the backend MCP server and the frontend UI.

Step 1: Install prerequisites on your system.

Step 2: Clone the project or obtain the MCP server bundle that contains the backend and frontend components.

Step 3: Create a Python virtual environment and install backend dependencies.

Step 4: Install frontend dependencies.

Step 5: Create and configure environment variables for LLM providers and server settings.

Step 6: Start the MCP server and the frontend interface.

Configuration and usage notes

You can configure multiple LLM providers in the environment and switch between them per request. Upload PDFs to a storage area, then use manual document selection or rely on automatic detection from your query text. The orchestrator analyzes your query, matches documents mentioned in the text, and routes tasks to internal and external agents before synthesizing a final answer.

Common workflow patterns include uploading a contract PDF, querying for specific clauses, and requesting cross-document comparisons across multiple uploaded files.

Additional sections

Configuration overview: You set environment variables for the backend and optionally for each provider. The system supports local Ollama, OpenAI, Anthropic, and Google providers, with per-provider models you can select in the UI.

Document management: Upload PDFs, extract text, and store documents for fast access. The system can automatically detect documents mentioned in queries, and you can also manually select documents to include in a query.

Troubleshooting tips cover common issues like ensuring the local provider is running, verifying API keys, and addressing port conflicts. If something seems off, validate environment variables and ensure the MCP server process is active.

Available tools

Upload PDF

Upload PDF documents to backend storage; PDFs are text-extracted and cached for fast access.

Document extraction

Automatic text extraction from PDFs using the backend processor to enable document-wide querying.

Manual document selection

Toggle which uploaded documents participate in a query via the UI sidebar.

Automatic document matching

Orchestrator identifies documents mentioned in the query text and matches them to available PDFs.

Agent orchestration

A multi-agent workflow that splits queries, assigns tasks to internal and external agents, and synthesizes results.

RESTful API endpoints

Standard endpoints to manage providers, models, documents, and orchestrated queries.

Built by

VeilStrat

AI signals for GTM teams