Hello fellow devs,
I am a 14 year old developer and I am incredibly excited to finally share this with you! For the past several weeks, I have been working day and night on a massive problem that was driving me absolutely crazy while vibe coding.
I use tools like Cursor, Trae, and Claude Code constantly to build my projects. But as my chat sessions grew longer, my development environments got incredibly slow, my token usage skyrocketed, and my AI assistants started completely forgetting the architectural decisions I made just a day prior!
I realized we are wasting up to 40 percent of our active context windows on repetitive boilerplate, verbose conversational filler, and identical duplicate code blocks.
I wanted a way to search, index, and compress my entire conversational history locally without paying for third-party vector databases or sending my private chat logs to external APIs.
So, at 14 years old, I built and launched Chronicle MCP! It is a high-performance local chat history archive connector built on the Model Context Protocol, and it is finally live!
How I Solved the Big Pain Points
- One-Click IDE Integration
I was so tired of copy-pasting code paths and manually editing hidden JSON configurations in
.cursoror.claude.json. It is incredibly annoying and prone to breaking! I wrote a smart, platform-agnostic automation system into the CLI. Now, you just run a single command:
chronicle add cursor
My script automatically scans your operating system (macOS, Windows, or Linux), finds your local installation of uvx, safely parses your editor settings, and injects the perfect stdio configuration directly! You can add it to Cursor, Trae, VS Code, Claude Code, or any emerging IDE instantly!
- The Conversation Splitter When you download your chat history from OpenAI or Anthropic, they hand you a giant, single, monolithic JSON file. You cannot feed that directly into an AI assistant! I built a powerful split engine that breaks that massive array into individual, neatly organized, and clean JSON files named after their actual historical titles. Just run:
chronicle split ~/Downloads/conversation.json --out ~/Desktop/MyChatLogs
- 25 Production-Ready Tools Once Chronicle connects via stdio, it exposes 25 independent local tools to your active LLM assistant. Your assistant can automatically query files, find related chats using a fast, zero-dependency local term frequency-inverse document frequency algorithm, extract action items/TODOs, and compile massive project briefs out of your historical context!
Get Started Right Now!
I wanted the installation to be as simple as humanly possible. If you have the uv package manager, you can install and run it globally right now:
# Install the tool globally
uv tool install chronicle-mcp-server
# Point Chronicle to your local chat logs directory
chronicle --chats-folder "~/Desktop/MyChatLogs"
# Inject the server directly into your favorite editor
chronicle add cursor
Once you run that, open up your editor's MCP settings panel. The connection will instantly light up solid green and start executing!
I put my absolute heart and soul into building this tool, testing the cross-platform path resolution, and getting the stdio channels to connect perfectly! It is fully open-source, and I would love to get your feedback, suggestions, or feature requests!
GitHub Repository: https://github.com/Leviathan0x0/Chronicle-MCP
PyPI Package: https://pypi.org/project/chronicle-mcp-server
Oh, and if you are wondering, the grammar of this post was improved using AI because I am not so good at it. Make sure to try my MCP server guys.
Top comments (0)