Supports deployment as an AWS Lambda function for automated daily data ingestion tasks to keep parliamentary data up-to-date.
Provides document storage and search capabilities for parliamentary data, with automatic embedding for semantic search over Hansard and Parliamentary Questions.
Utilizes Azure OpenAI for embedding documents to enable semantic search functionality across parliamentary data.
Parliament MCP Server
An MCP server that roughly maps onto a subset of https://developer.parliament.uk/, as well as offering additional semantic search capabilities.
Architecture
This project provides:
- MCP Server: FastMCP-based server
- Python package: A small python package for querying and loading parliamentary data from https://developer.parliament.uk/ into Elasticsearch
- Elasticsearch: Document storage and search engine, setup to automatically embed documents and allow semantic search over Hansard and Parliamentary Questions data.
- Claude Integration: Connect to Claude Desktop via
mcp-remote
proxy
Features
MCP Tools Available
The MCP Server exposes 11 tools to assist in parliamentary research:
search_constituency
- Search for constituencies by name or get comprehensive constituency details by IDget_election_results
- Get election results for constituencies or specific memberssearch_members
- Search for members of the Commons or Lords by various criteriaget_detailed_member_information
- Get comprehensive member information including biography, contact, interests, and voting recordget_state_of_the_parties
- Get state of the parties for a house on a specific dateget_government_posts
- Get exhaustive list of all government posts and their current holdersget_opposition_posts
- Get exhaustive list of all opposition posts and their current holdersget_departments
- Get reference data for government departmentssearch_parliamentary_questions
- Search Parliamentary Written Questions (PQs) by topic, date, party, or membersearch_debates
- Search through debate titles to find relevant debatessearch_contributions
- Search Hansard parliamentary records for actual spoken contributions during debates
Quick Start (local elasticsearch)
You will need
- Docker and Docker Compose
- Node.js (for mcp-remote)
- Claude Desktop (or another MCP client)
- Azure OpenAI account with API access
Create a .env
file by copying the .env.example
in the project root and replace the necessary variables.
After configuring your .env
file, run the following command for a one shot setup of the MCP server and the Elasticsearch database with some example data from June 2025.
Once this is run, you can connect to the MCP server using this config
Manual Setup
1. Clone, setup environment, and start services
Note on Elasticsearch Configuration
The system supports two methods for connecting to Elasticsearch:
- Local/Self-hosted: Use
ELASTICSEARCH_HOST
,ELASTICSEARCH_PORT
, andELASTICSEARCH_SCHEME
- Elasticsearch Cloud: Use
ELASTICSEARCH_CLOUD_ID
andELASTICSEARCH_API_KEY
(automatically configures host, port, and scheme)
The services will be available at:
- MCP Server:
http://localhost:8080/mcp/
- Elasticsearch:
http://localhost:9200
3. Initialize Elasticsearch and Load Data
2. Install mcp-remote
3. Configure Claude Desktop
Add the following to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
You can also generate this configuration automatically:
4. Restart Claude Desktop
Claude should now have access to the Parliament MCP tools.
Development
Prerequisites for Local Development
- Python >= 3.12
- uv (Python package manager)
- Docker and Docker Compose
- Node.js (for mcp-remote)
Local Development Setup
- Install uv (if not already installed):
- Clone and set up the project:
- Available Make commands:
- Run the MCP server locally:
Project Structure
CLI Commands
The project includes a CLI for data management:
Data Structure
The system works with two main types of parliamentary documents:
Parliamentary Questions (Index: parliamentary_questions
):
- Written Questions with semantic search on question and answer text
- Member information for asking and answering members
- Date, reference numbers, and department details
Hansard Contributions (Index: hansard_contributions
):
- Spoken contributions from parliamentary debates
- Semantic search on full contribution text
- Speaker information and debate context
- House (Commons/Lords) and sitting date
Data Loading Process:
- Fetch from Parliamentary APIs (Hansard API, Parliamentary Questions API)
- Transform into structured models with computed fields
- Embed using Azure OpenAI for semantic search
- Index into Elasticsearch with proper mappings
Data is loaded automatically from official Parliamentary APIs - no manual document creation needed.
Daily Data Ingestion
To keep the data in Elasticsearch up-to-date, a daily ingestion mechanism is provided. This loads the last two days of data from both hansard
and parliamentary-questions
sources.
To run the daily ingestion manually:
This runs the equivalent of:
For automated daily ingestion, you can use a cron job. Here are examples for standard cron and AWS EventBridge cron.
Standard Cron
This cron job will run every day at 4am.
AWS EventBridge Cron
This AWS EventBridge cron expression will run every day at 4am UTC.
Notes on AWS Lambda Deployment for Daily ingestion
A docker based AWS lambda image is provided to run daily ingestion tasks.
1. Build the Lambda Container Image
Build the Docker image using the provided Makefile
target:
This will create a Docker image named parliament-mcp-lambda:latest
.
2. Test the Lambda locally
You can test the Lambda function locally using the AWS Lambda Runtime Interface Emulator (RIE), which is included in the base image.
Prerequisites:
- Your local Elasticsearch container must be running (
docker compose up -d elasticsearch
). - The Lambda container image must be built (
make docker_build_lambda
).
Run the container:
A convenient way to provide the necessary environment variables is to use the --env-file
flag with your .env
file. You still need to override the ELASTICSEARCH_HOST
to ensure the container can connect to the service running on your local machine.
Trigger the function:
Open a new terminal and run the following curl
command to send a test invocation. The from_date
and to_date
parameters are optional. If not provided, it will default to loading everything from the last 2 days.
- Configure the Lambda in AWS
With the image pushed to ECR, you can create the Lambda with the following configurations
- Use
ELASTICSEARCH_HOST
,ELASTICSEARCH_PORT
, andELASTICSEARCH_SCHEME
to point to your elasticsearch cluster - Increase the default timeout to ~10 minutes to ensure the ingestion has enough time to complete
- Use AWS's flavour of cron to schedule the task for ~4am every day -
cron(0 4 * * ? *)
- Remember to increase the default timeout to ~10 minutes to ensure the ingestion has enough time to complete.
Usage Examples
Once connected to Claude, you can use natural language queries like:
Parliamentary Questions:
- "Search for parliamentary questions about climate change policy"
- "Find questions asked by Conservative MPs about healthcare funding"
- "Show me recent questions about education from the last week"
Hansard Contributions:
- "Search for contributions about the budget debate"
- "Find speeches by Keir Starmer about economic policy"
- "Show me debates from the House of Lords about immigration"
Member Information:
- "Get detailed information about the MP for Birmingham Edgbaston"
- "Search for Labour MPs elected in 2024"
- "Find constituency information for postcode SW1A 0AA"
Parliamentary Structure:
- "Show me the current government ministers"
- "Get the state of the parties in the House of Commons"
- "List all opposition shadow cabinet positions"
General Queries:
- "Search for debates about artificial intelligence regulation"
- "Find election results for marginal constituencies"
- "Show me government departments and their responsibilities"
Logs and Debugging
View server logs:
Enable debug mode in Claude config by adding --debug
flag.
Check Elasticsearch status:
Troubleshooting
Common Issues
MCP Connection Issues
- Ensure MCP server is running on port 8080
- The MCP server runs on
/{MCP_ROOT_PATH}/mcp
, not/MCP_ROOT_PATH
- Verify Claude Desktop configuration is correct
Data Loading Failures
- Check Azure OpenAI credentials in
.env
file - Ensure Elasticsearch is running and accessible
- Verify network connectivity to Parliamentary APIs
- Use
--ll DEBUG
flag for detailed logging
Elasticsearch issues
- Verify inference endpoints are created:
parliament-mcp init-elasticsearch
- Use https://elasticvue.com/ to inspect the Elasticsearch instance
Contributing
Contributions are welcome! Please see our Contributing Guide for details on how to get started.
License
MIT License - see LICENSE file for details
This server cannot be installed
An MCP server that provides tools for parliamentary research by allowing users to search constituencies, elections, members, government posts, debates, and Hansard records, with additional semantic search capabilities over parliamentary data.
Related MCP Servers
- AsecurityFlicenseAqualityAn MCP server that enables users to search for government grants based on keywords and view detailed information including funding, eligibility, and deadlines.Last updated -1JavaScript
- AsecurityFlicenseAqualityA MCP server that allows searching for files in the filesystem based on path fragments, returning file metadata including name, path, size, and creation date.Last updated -1Python
- -securityFlicense-qualityAn MCP server that enables searching and retrieving content from Confluence documentation systems, providing capabilities for both document searches and full page content retrieval.Last updated -Python
- -securityAlicense-qualityStores metadata for MCP servers and provides smart search capabilities, allowing users to find appropriate MCP servers for their queries and route requests to the most suitable server.Last updated -9PythonMIT License