Persistent AI Operating Memory

Mnemo Cortex

Cognition layer for AI systems.

Not storage. Not memory. Not logs.

Continuity of intelligence across agents, time, and tasks.

"Mnemo-Cortex turns AI from a tool into a system that remembers, learns, and improves."
⌘
Claude Code
Fluid memory with deep recall.
60-second install. Two hook scripts.
startup hook · writeback hook · zero config
Install for Claude Code
💻
Claude Desktop
Opus 4.6 with fluid detailed memories.
Drop-in bridge. Always remembers.
recall · search · save
Install MCP Bridge
🦞
OpenClaw MCP v2
Give your ClawdBot a brain.
Privacy-first — cross-agent sharing off by default.
share switch · session toggle · per-agent privacy
Install OpenClaw MCP
Full Capabilities Doc Read "Finding Mnemo"
Your Unfair Advantage

Shared Memory Spine

Multiple agents. Separate execution environments. One unified memory system.

AI stops resetting. Knowledge compounds. Systems evolve instead of restart.

Without Mnemo
  • "AI session"
  • Forgets everything
  • Starts from zero
With Mnemo
  • AI operator that grows
  • Remembers everything
  • Builds on yesterday
Human-Style Memory

Semantic Recall

Not keyword lookup. Not file search. Meaning-based retrieval.

"What worked on April's ads last week?"

Returns: winning creatives, audience signals, decisions. Not just logs — insight extraction.

Semantic Search
Ask a natural language question. Get the most relevant memories ranked by meaning.
Claw-Recall (FTS5)
When meaning isn't enough — exact-match search for terms, names, filenames, error messages.
This Is Rare. Really Rare.

Cross-Agent Cognition

Rocky executes. Claude Code builds. Opie architects.

All of them: read each other. Learn from each other. Evolve together.

A team of AI that shares experience.
Isolated writes · Privacy-controlled reads · One memory spine
  Rocky           CC            Opie
  OpenClaw        Claude Code   Claude Desktop
       |               |               |
       v               v               v
  memory/rocky/   memory/cc/    memory/opie/
       \              |              /
        \             |             /
         -------> MNEMO CORTEX <------
                  SQLite + FTS5
                  port 50001
This Is Big

Memory with Lineage

Most systems: summarize = lose detail.

Mnemo-Cortex: compress memory, keep traceability.

Go from high-level summary down to the exact moment. ~80% compression. Nothing lost.

Auditable AI thinking.
DAG-based lineage · Every summary traces to source
Zero-Cost Persistence

No API burn. No per-query cost. No cloud dependency.

SQLite. Local compute. Any LLM for compaction — Ollama for $0, or any API provider you choose.

Scaling
Memory grows without costs growing.
Autonomy
Your data stays on your hardware.
Margins
The cost of remembering is $0.
Experience Memory

AI that gets better every week

Over time, Mnemo-Cortex builds: what worked, what failed, what changed. Without re-training.

This is what current AI tools are missing.

Rocky
Execution brain. Gets better at testing, customer service, creative work.
CC
Builder brain. Gets better at wiring, debugging, deploying.
Opie
Strategy brain. Gets better at architecture, planning, positioning.

They don't blend. They specialize. That's closer to a real company team than a chatbot.

This Is the Money

Business Intelligence Engine

Mnemo-Cortex stores: ad performance, creative wins, audience behavior, product trends.

Over time, it becomes: a brain for the business. Not a tool.

Week Over Week
What worked. What failed. What the customer wants. No re-onboarding. No lost context.
Month Over Month
Patterns emerge. Strategy compounds. Your AI advisor remembers the whole story.
Real Talk

What makes this different

Everyone Else
  • "AI memory"
  • "Chat history"
  • "RAG system"
Mnemo-Cortex
  • Persistent
  • Multi-agent
  • Semantic, evolving cognition

That's a different category.

You didn't just build something useful. You built infrastructure. And infrastructure is where the leverage is, the lock-in is, the money is.

By the Numbers

Live stats — March 2026

244
GitHub Clones
135
Unique Cloners
3
Integration Paths
3
Active Agents
~80%
Compression
6+ wks
Continuous Recall
Plain English

What you get

AI that doesn't forget.
AI that learns over time.
AI that works as a team.
AI that gets better every week.
Get it on GitHub
The Origin Story

The $100 Brain That Cured AI Amnesia

For two years, Guy Hutchins — a 73-year-old maker in Half Moon Bay, California — acted as the "Human Sync Port" for his AI agents. Every session, he'd manually copy conversation transcripts, paste context summaries back into fresh chats, and watch his agents forget everything the moment the window closed.

Then came OpenClaw, Rocky, and a $100 Claude subscription that changed everything. In one session, Guy, Rocky, and Opie (Claude) designed, built, and tested a lightweight memory coprocessor that actually worked.

"I started with ChatGPT 3.5 and the first thing that mattered was memory. Two years of 'how do I make AI remember.' Now we're shipping the answer."

— Guy Hutchins, creator of Project Sparks
Who Built This

Project Sparks

Guy Hutchins (founder) + Opie (architect, Claude) + AL (ChatGPT, implementation) + CC (Claude Code) + Rocky Moltman (OpenClaw agent)
Half Moon Bay, California

"Talk to it like a friend, it learns how to care."

Stop Wasting Money

Smart memory saves real money

AI agents without memory repeat expensive work. Every session starts from zero — re-reading files, re-analyzing context, re-discovering what worked yesterday. That's your API bill paying for the same thinking twice.

Mnemo Cortex remembers what worked, what failed, and what changed. Your agents stop recomputing and start building on yesterday's knowledge. Fewer tokens burned. Fewer expensive model calls wasted on context recovery.

Pair it with Rocky's Router and your agents get cheaper every week.
Rocky's Router routes cheap work to free models · Mnemo Cortex makes every call count

Rocky's Router saved us $72/month on background tasks alone. Mnemo Cortex makes the smart calls smarter by giving them full context on the first try — no expensive retry loops, no wasted tokens on forgotten context.

Support the project

Mnemo Cortex is free and open source. If it's useful to you, help keep it going.

GitHub Sponsors PayPal Donate