Kokkai Minutes MCP Agent
# Kokkai Minutes MCP Agent
A Model Context Protocol (MCP) agent that provides a structured interface to the Japanese National Diet Library's Kokkai Minutes (parliamentary proceedings) API. This agent allows AI models to search and retrieve parliamentary proceedings through a standardized interface.
🚀 **Already Deployed**: The agent is live and ready to use at https://kokkaikaigiroku-mcp.egeg.workers.dev/mcp
## Table of Contents
- [Example Questions](#example-questions-for-llms)
- [Installation](#installation)
- [Development](#development)
- [Deployment](#deployment)
- [API Functions](#api-functions)
- [Environment](#environment)
- [Connect Claude Desktop](#connect-claude-desktop-to-your-mcp-server)
- [Contributing](#contributing)
## Example Questions for LLMs
Here are some example questions you can ask AI models (like Claude) that use this agent:
### Policy Research
- "Give me the list of politicians that talked about problems with 仮放免 (provisional release) in the last year"
- "What were the main arguments made about 入管法改正 (immigration law reform) in recent Diet sessions?"
- "Summarize discussions about 待機児童 (childcare waiting lists) from 2020 to present"
### Political Analysis
- "How has the government's stance on 原発再稼働 (nuclear power plant restart) evolved over time?"
- "Find statements by the Prime Minister about インバウンド観光 (inbound tourism) after COVID-19"
- "Compare different party positions on 消費税 (consumption tax) in budget committee meetings"
### Historical Research
- "Find debates about 日米安保条約 (US-Japan Security Treaty) from its original ratification period"
- "How did the Diet discuss 少子化対策 (measures against declining birthrate) when it first became a social issue?"
## API Functions
The agent provides three main functions:
1. `searchMeetingList` - Search for meetings with basic information
2. `searchMeetingsWithSpeechText` - Search for meetings including full speech texts
3. `searchSpeeches` - Search for specific speeches
Each function supports various parameters including:
- Date range (`from`, `until`)
- House selection (`nameOfHouse`)
- Speaker information (`speaker`, `speakerPosition`, `speakerGroup`)
- Content search (`any`)
- Session numbers (`sessionFrom`, `sessionTo`)
- And more
For detailed parameter information, please refer to the source code documentation.
## Environment
- Node.js
- TypeScript
- Cloudflare Workers
- Model Context Protocol SDK
## Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
## Connect Claude Desktop to your MCP server
You can connect to the MCP server from local MCP clients using the [mcp-remote proxy](https://www.npmjs.com/package/mcp-remote).
To connect from Claude Desktop, follow [Anthropic's Quickstart](https://modelcontextprotocol.io/quickstart/user) and within Claude Desktop go to Settings > Developer > Edit Config.
Update with this configuration:
```
{
"mcpServers": {
"kokkaikaigiroku": {
"command": "npx",
"args": [
"mcp-remote",
"https://kokkaikaigiroku-mcp.egeg.workers.dev/mcp" // Use this for the deployed version
// or "http://localhost:8787/mcp" for local development
]
}
}
}
```
Restart Claude and you should see the Kokkai Minutes search tools become available. TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose with no ambiguity: searchMeetingList retrieves meeting metadata without speech text, searchMeetingsWithSpeechText returns meetings with full speech text, and searchSpeeches focuses on individual speeches with meeting context. The descriptions explicitly differentiate their scopes and output formats, making misselection unlikely.
All tool names follow a consistent verb_noun pattern using camelCase (searchMeetingList, searchMeetingsWithSpeechText, searchSpeeches), with 'search' as the uniform verb prefix. This predictable naming scheme enhances readability and agent usability without deviations.
With only 3 tools, the count feels borderline thin for a parliamentary minutes domain, potentially limiting advanced operations like filtering by date or speaker. However, it covers core search functionalities adequately, avoiding bloat but leaving room for expansion to improve workflow completeness.
The tool set provides comprehensive search coverage for Kokkai minutes, including metadata, full meeting texts, and individual speeches, with no dead ends. Minor gaps exist, such as lack of update/delete operations (irrelevant for read-only data) or advanced filtering options, but agents can work around these with the available tools.