kevin-hs-sohn

kevin-hs-sohn / hipocampus

Public

Drop-in memory harness for AI agents — 3-tier memory, compaction tree, hybrid search. One command to set up. Works with Claude Code and OpenClaw.

agent-memory ai-agent ai-memory claude-code compaction-tree
10
1
100% credibility
Found Mar 17, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

Hipocampus is a simple file-based system that adds persistent memory to AI agents in Claude Code and OpenClaw projects, helping them recall past work, tasks, and knowledge across sessions with automatic organization.

How It Works

1
🔍 Discover Hipocampus

You learn about a handy tool that gives your AI helper a lasting memory, so it remembers past chats and work without starting over each time.

2
📁 Open your AI project

Head to the folder where you're teaming up with your AI assistant to build or create things.

3
🚀 Set up memory home instantly

With one easy action, you create cozy notebooks for short-term notes, long-term facts, task lists, and a smart index that organizes everything automatically.

4
🤖 Chat with your upgraded AI

Your assistant now greets you by loading its memory summary, knowing recent tasks, decisions, and what it has learned before.

5
📝 Work and save progress

As you tackle projects together, it jots down findings, updates plans, and files away lessons into daily summaries that build over time.

6
💡 Recall details effortlessly

Need info from weeks ago? It checks its quick index, grabs just what's relevant, and shares it without wasting time or repeating old work.

🎉 Reliable growing partner

Your AI becomes a smart companion that remembers everything important, saving you time and making projects smoother and more fun.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is hipocampus?

Hipocampus is a JavaScript drop-in memory harness for AI agents, mimicking the hippocampus gehirn's role in human memory—storing and retrieving info efficiently. Run `npx hipocampus init` in a Claude Code or OpenClaw project, and it sets up a 3-tier memory system with compaction tree and optional hybrid search via qmd, all using plain files, no servers needed. Developers get persistent agent recall across sessions at low token cost, avoiding blind re-exploration of old work.

Why is it gaining traction?

Unlike ad-hoc memory files or heavy RAG setups, hipocampus drops in like Adyen drop-in GitHub integrations or Braintree web drop-in—zero infra, instant agent skills, and smart compaction that scales months without overflow. The hook is ROOT index awareness: agents know what they know in 3K tokens, enabling targeted retrieval over wasteful full-context dumps. Hybrid search and proactive flushes beat drop memory cache pitfalls, blending BM25 with vectors for discovery.

Who should use this?

AI agent builders on Claude Code or OpenClaw tackling long-term projects, like automating codebases or research pipelines, where session amnesia kills productivity. Solo devs or small teams training hippocampus funktion via repeated tasks, not one-offs needing git drop commit resets.

Verdict

Promising early experiment (10 stars, 1.0% credibility) with thorough docs but zero tests—try for niche drop-in github memory if you're deep in those platforms, but watch for maturity before production. Solid for hacking agent persistence now.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.