contract-risk-analyzer
Analyzes financial contract PDFs to extract clauses, flag risk terms, and compare contract versions, producing structured risk briefs for legal and risk teams.
README
contract-risk-analyzer
Contract Risk Analyzer is a FastMCP server + LangGraph workflow that ingests financial contract PDFs, extracts key clauses and obligations, flags known risk terms with severity, compares contract versions, and synthesizes everything into a structured risk brief for lawyers, risk teams, and operators who need fast, explainable contract triage.
The MCP tools accept either a local file_path or a remote pdf_url. For hosted deployments such as Railway, use pdf_url so the server can download the PDF into temporary storage before analysis.
Architecture (high level)
PDF
|
v
FastMCP_Server
|
+--> extract_clauses
+--> flag_risk_terms
+--> summarize_obligations
+--> compare_contracts
|
v
LangGraph_Agent (orchestrates tools)
|
v
RiskBrief (Pydantic structured output)
Setup (local)
cd contract-risk-analyzer
cp .env.example .env
source .venv/bin/activate # if you already created the project virtualenv
pip install -e ".[dev]"
python -m contract_risk_analyzer.server
- MCP endpoint:
http://localhost:8000/mcp - Health check:
http://localhost:8000/health - The
.envfile must containOPENAI_API_KEY.
Connecting from Claude Desktop (MCP client)
Option A: Run as a local STDIO server (Claude Desktop spawns it)
In Claude Desktop, add an MCP server entry similar to:
{
"mcpServers": {
"contract-risk-analyzer": {
"command": "python",
"args": ["-m", "contract_risk_analyzer.server"],
"env": {
"OPENAI_API_KEY": "YOUR_KEY_HERE"
}
}
}
}
Option B: Connect to the local HTTP server
If you run the server yourself (python -m contract_risk_analyzer.server), bridge Claude Desktop to the local MCP HTTP endpoint with mcp-remote:
{
"mcpServers": {
"contract-risk-analyzer": {
"command": "npx",
"args": [
"mcp-remote",
"http://127.0.0.1:8000/mcp",
"--allow-http"
]
}
}
}
Option C: Connect to the deployed Railway server
The deployed server is available at:
- Health check:
https://contract-risk-analyzer-production-410a.up.railway.app/health - MCP endpoint:
https://contract-risk-analyzer-production-410a.up.railway.app/mcp
Claude Desktop config:
{
"mcpServers": {
"contract-risk-analyzer": {
"command": "npx",
"args": [
"mcp-remote",
"https://contract-risk-analyzer-production-410a.up.railway.app/mcp"
]
}
}
}
Example tool calls
Input source rules
For single-contract tools, provide exactly one of:
{
"file_path": "/app/samples/contract.pdf"
}
or:
{
"pdf_url": "https://example.com/contracts/contract.pdf"
}
For compare_contracts, provide exactly one source for each side:
{
"pdf_url_a": "https://example.com/contracts/v1.pdf",
"pdf_url_b": "https://example.com/contracts/v2.pdf"
}
Remote PDFs are downloaded to temporary storage, capped at 50 MB per PDF, and deleted after each tool call.
extract_clauses
Input:
{
"pdf_url": "https://example.com/contracts/isda.pdf",
"clause_type": "termination events"
}
Sample output:
[
{
"section_name": "ARTICLE_VII TERMINATION",
"clause_type": "termination events",
"raw_text": "…",
"plain_english": "…",
"page_references": [12, 13]
}
]
flag_risk_terms
Input:
{ "pdf_url": "https://example.com/contracts/isda.pdf" }
Sample output:
[
{
"term": "cross-default",
"context": "…",
"risk_explanation": "…",
"severity": "high",
"page_reference": 9
}
]
summarize_obligations
Input:
{ "pdf_url": "https://example.com/contracts/isda.pdf" }
Sample output:
[
{
"party": "Borrower",
"obligations": ["Deliver monthly financial statements…"],
"key_deadlines": ["Within 30 days after month-end…"],
"conditions": ["So long as no Event of Default has occurred…"]
}
]
compare_contracts
Input:
{
"pdf_url_a": "https://example.com/contracts/v1.pdf",
"pdf_url_b": "https://example.com/contracts/v2.pdf"
}
Sample output:
{
"added_clauses": ["New collateral top-up requirement…"],
"removed_clauses": ["Removed cure period for payment default…"],
"materially_changed_clauses": [
{
"section_name": "ARTICLE_IV EVENTS_OF_DEFAULT",
"change_summary": "Acceleration now triggers immediately…",
"risk_note": "Increases lender leverage; reduces borrower flexibility."
}
],
"risk_delta": "Overall risk increased for Borrower due to tighter default/acceleration terms."
}
Deployment (Railway)
- Build: Railway will build the container from
Dockerfile. - Run: The container runs
python -m contract_risk_analyzer.serverand binds to$PORT(default8000). - Health check:
GET /healthreturns{"status":"ok"}. - Environment: Set
OPENAI_API_KEYin Railway service variables. - Current deployment:
https://contract-risk-analyzer-production-410a.up.railway.app.
Tech stack
- FastMCP
- OpenAI GPT-4o (via
openaiSDK) - PyMuPDF (
pymupdf) - Pydantic
- LangGraph
- Docker
- Railway
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.