Story Collecting Agent is an AI-powered tool built using Model Context Protocol (MCP) with Claude Desktop (via stdio and HTTP).
It collects data points related to user NLP queries, integrates with Google Search to fetch relevant stories, and organizes them into a classified Excel format stored locally.
This project demonstrates how MCP-based agents can be integrated with search engines and structured storage to build a knowledge pipeline.
- Collects and processes user NLP queries
- Fetches relevant information from Google Search
- Stores results in classified Excel spreadsheets for easy analysis
- Built with MCP for seamless integration with Claude Desktop
- Dependency management using uv (faster, modern alternative to pip/venv)
- Python 3.10+
- uv installed (for dependency management)
- Claude Desktop with MCP enabled
Clone the repository:
git clone https://github.com/AswinKumar1/Story_collecting_agent.git
cd Story_collecting_agentInstall dependencies using uv:
uv syncRun the main agent:
uv run python main.py(Optional) Run debug/testing scripts:
uv run python debug.py
uv run python test.py| File/Folder | Description |
|---|---|
main.py |
Entry point for running the Story Collecting Agent |
debug.py |
Debugging utilities |
test.py |
Test scripts for validation |
classifier_learning.db |
Database storing classification logic/data |
us_freedom_stories.xlsx |
Example dataset of collected stories |
*.xlsx files |
Output files containing classified stories |
pyproject.toml, uv.lock |
Project metadata and uv dependency lock file |
- User enters a query (NLP text input).
- Agent connects via Claude Desktop MCP (stdio + HTTP).
- Query is enriched and sent to Google Search.
- Relevant stories/data are retrieved and classified.
- Results are saved into structured Excel files locally.
Contributions are welcome! Please open an issue or submit a pull request if you’d like to add features, improve docs, or fix bugs.
This project is licensed under the MIT License.