Hello, I'm Rijul. I'm building git-lrc, a micro AI code reviewer that runs on every commit. It's free and source-available on GitHub. Star git-lrc to help more developers discover the project. Do give it a try and share your feedback
As I use AI coding agents to work on my repositories, I often find myself opening new sessions for different tasks.
The problem is that every new session starts with the same challenge.
The agent may have to rediscover the codebase.
It may search for files, follow imports, look through related code, and try to understand the project structure all over again.
This gets annoying pretty quickly.
It also means spending additional time, tokens, and context on work that has already been done in previous sessions.
So I started looking for a way to solve this problem, since I imagine many people using coding agents run into the same issue.
That's when I came across codebase-memory-mcp.
What Is codebase-memory-mcp?
You can find the repository here:
https://github.com/DeusData/codebase-memory-mcp
The basic idea is to create a persistent knowledge graph of your codebase that an AI agent can query.
Think about a textbook.
A textbook usually has an index that tells you where to find specific topics instead of requiring you to read the entire book to locate something.
codebase-memory-mcp applies a similar idea to your codebase.
Instead of making the agent repeatedly explore files and related code to figure out where something is, it can query the indexed representation of the codebase and use that information to find the relevant parts.
This gives the agent a better starting point when working on a repository.
Rather than repeatedly asking:
Where is this functionality implemented?
Which files are connected to this component?
Where is this function used?
The agent can use the codebase index to navigate the repository more efficiently.
Installing codebase-memory-mcp
You can follow the installation instructions in the repository's README.
For Linux, you can install it with:
curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash
Once the installation is complete, you can run:
codebase-memory-mcp install
This configures the MCP server for a number of supported AI coding agents, including OpenCode and Claude Code.
After that, you need to index your project before the agent can query the indexed codebase. The project's README recommends restarting your coding agent and asking it to "Index this project".
Wrapping Up
When we start using AI agents for software development, there are new limitations that we don't encounter in the same way with traditional development.
One of them is the amount of time and context an agent can spend rediscovering a codebase.
Tools like codebase-memory-mcp are interesting because they try to solve that problem by giving agents a persistent representation of the codebase that they can query instead of starting from scratch every time.
As we adapt to agent-based development, understanding these limitations and finding ways to work around them can make the overall development process much smoother.
AI agents write code fast. They also silently remove logic, change behavior, and introduce bugs -- without telling you. You often find out in production.
git-lrc fixes this. It hooks into git commit and reviews every diff before it lands. 60-second setup. Completely free.
Any feedback or contributors are welcome! It's online, source-available, and ready for anyone to use.
Give it a ⭐ star on Github
Top comments (0)