MCP server · Developer tools
True Memory Fragments MCP server
by kyle641320
Lets your AI check whether its memory of your code is still up to date before it acts on it.
True Memory Fragments (TMF for short) is a helper for AI coding assistants that work on your code across many sessions. It keeps track of what the AI learned about your code and checks whether that knowledge is still true after the code changes. It is handy for developers who use an AI coding agent on a codebase that keeps moving.
What is an MCP server? The 30-second version
On its own, your AI can only chat with you. An MCP server is a small helper program that gives your AI a new skill or a connection to something else. This one connects your AI to a memory store tied to your code, so the AI can ask whether a fact it remembers still matches the current files. When the code has changed, the helper tells the AI to stop and reread the source instead of trusting old notes.
What this MCP server does
You ask your AI something about your code, and the AI asks this helper what it knows. The helper looks at the claims it stored earlier, like "function A calls function B", and checks whether the files those claims came from have changed since. If a claim is still fresh, the AI gets a small, bounded picture of related code. If a claim is stale or unknown, the helper refuses to hand it over and instead points the AI back to the exact file to reread. That way your AI does not quietly reuse old assumptions after you have changed the code.
Click to zoomWhat you can do with it
- Check whether remembered facts about your code still match the current files
- Spot stale call chains and dependency assumptions before your AI acts on them
- Get source anchors that point your AI back to the file it should reread
- Explore a bounded view of related code around a claim
- Mark async handoffs like publish and subscribe relationships
- Stop graph exploration when a claim is stale or unknown
- Run a quick demo that shows stale-claim blocking in action
Try asking your AI
- “Is your memory of how orders flow through this code still accurate after my last commit”
- “Which of your stored claims about this file are now stale”
- “Show me the code around the function that calls sendEmail and tell me if it changed”
- “Before you answer, check whether the source for that claim is still fresh”
What it gives back to you
You get answers in the chat that either confirm a claim is still fresh or flag it as stale or unknown. When something is stale, you get a short note plus a pointer to the exact file that needs rereading. Fresh results come back as a small, bounded set of related code facts, not a giant dump. The helper does not change your code; it only reports and points.
Before you start
What you need
- Python 3.10 or newer
- Git installed
- A source checkout of the project if you want to run the demo
- An AI coding agent that supports MCP, such as Claude Code, if you want the reflex hook
Good to know
It is a developer preview, so behavior depends on your setup, and "fresh" only means the file still matches, not that the fact is correct.
Install it with your AI
Add True Memory Fragments MCP server to your AI, no technical skills needed
You don't install anything by hand. You copy one prompt, paste it into an AI that can work on your computer, and it checks, installs and connects the server for you, asking you when it needs something.
Sign in to get the install prompt
Members get a ready-made prompt that lets the Claude desktop app check True Memory Fragments MCP server, install it and connect it for them, step by step. You don't need any technical skills: you copy, paste and answer a few questions. Your connected AI can also find and install any of the 4,066 MCP servers here for you.
Who it's for
Developers and teams who use AI coding agents on codebases that change often and want to avoid the AI acting on outdated assumptions.
