Skip to content
BSWEN
Blog

How to Add Memory to Claude Code: Best Repos for Context Persistence in 2026

Cowrie
Cowrie
Dev @ Bswen

I kept re-explaining my project to Claude Code every single session. It was frustrating.

Every time I started a new Claude Code session, I had to:

  • Re-describe my project architecture
  • Re-explain my coding preferences
  • Re-share the same context about past decisions

I thought there had to be a better way. Turns out, there is.

The Problem: Claude Code Has No Memory

Claude Code starts fresh with each session. No memory of:

  • What we built last week
  • Why we made certain architectural decisions
  • My coding style preferences
  • Project-specific knowledge we already covered

This meant I was spending 10-15 minutes per session just setting context. Multiply that by multiple sessions per day, and that’s a lot of wasted time.

The Solution: Persistent Memory Repos

I found two GitHub repositories that solve this problem:

graph LR
    A[Claude Code Session] --> B{Memory Type?}
    B -->|Simple Recall| C[Claude Mem]
    B -->|Complex Docs| D[LightRAG]
    C --> E[Session Persistence]
    D --> F[Knowledge Graphs]
    E --> G[Remember Context]
    F --> G

Claude Mem: Simple Session Memory

Claude Mem (thedotmack/claude-mem) gives Claude Code basic memory capabilities. It stores context between sessions so you don’t have to repeat yourself.

terminal
# Clone Claude Mem
git clone https://github.com/thedotmack/claude-mem


# Navigate to the directory
cd claude-mem


# Follow the setup instructions in README

What it does:

  • Remembers project context across sessions
  • Stores coding preferences
  • Recalls past decisions

LightRAG: Advanced Document Retrieval

LightRAG (hkuds/lightrag) from the University of Hong Kong is a lightweight RAG (Retrieval-Augmented Generation) framework. It’s more powerful than simple memory:

terminal
# Clone LightRAG
git clone https://github.com/hkuds/lightrag


# Install dependencies
cd lightrag
pip install -r requirements.txt

What it does:

  • Builds knowledge graphs from your documents
  • Enables semantic search across large codebases
  • Provides context-aware retrieval for complex queries

When to Use Which?

decision-guide.txt
+------------------+------------------+
| Use Case         | Recommended Tool |
+------------------+------------------+
| Simple recall    | Claude Mem       |
| Large codebase   | LightRAG         |
| Multiple docs    | LightRAG         |
| Quick setup      | Claude Mem       |
| Knowledge graphs | LightRAG         |
| Session-based    | Claude Mem       |
+------------------+------------------+

I started with Claude Mem for my personal projects. It’s simpler and does the job for most use cases.

For my work codebase (thousands of files), LightRAG shines. The knowledge graph approach means I can ask questions like “How does the auth system work?” and get accurate, context-aware answers.

Common Mistakes to Avoid

Mistake 1: Assuming Claude remembers

Nope. Each session starts blank. You need explicit memory tools.

Mistake 2: Not using RAG for large codebases

For projects with lots of documentation or files, simple memory isn’t enough. LightRAG’s retrieval capabilities are essential.

Mistake 3: Over-complicating setup

Start simple. Claude Mem takes minutes to set up. Only move to LightRAG if you need advanced features.

How I Use It Now

My workflow has changed significantly:

  1. Project setup: I configure memory storage once
  2. First session: I explain my project (just once)
  3. Future sessions: Claude remembers everything
time-saved.txt
Before: 15 min context-setting per session
After:  30 seconds (just load the memory)


Daily savings: ~45 minutes
Weekly savings: ~4 hours

Key Takeaways

  • Claude Code needs external tools for persistent memory
  • Claude Mem is best for simple session-based recall
  • LightRAG excels with large codebases and documentation
  • Start simple, scale up when needed

Persistent memory transforms Claude Code from a session-based assistant into a long-term coding partner. The initial setup is worth the time you’ll save across countless future sessions.

Final Words + More Resources

My intention with this article was to help others share my knowledge and experience. If you want to contact me, you can contact by email: Email me

Here are also the most important links from this article along with some further resources that will help you in this scope:

Oh, and if you found these resources useful, don’t forget to support me by starring the repo on GitHub!

Comments

Tags: