MemPalace Toolkit

Enterprise deployment automation for MemPalace

Multi-project wiring · Cross-session memory · Team deployment templates

Enterprise-grade memory for teams using multiple AI assistants

---

What It Is

MemPalace Toolkit extends the official MemPalace (50K+ stars) with enterprise-grade deployment automation:

• Multi-project wiring — 20+ projects share a single unified memory palace
• Cross-session memory — Claude, Codex, Gemini, Qwen all save to the same palace
• Template-based deployment — One command wires any new project
• Team coordination — Shared knowledge across all AI assistants

Built on top of the amazing MemPalace by Milla Jovovich.

---

Real-World Usage

✨ Your Ecosystem - Production deployment:

• 20+ projects wired with unified memory
• 5 AI sessions (Claude/Codex/Gemini/Qwen/claude-fcc) sharing single palace
• Automated hooks saving memory on every session stop/compact

---

Quick Start

Install

pip install mempalace-toolkit

Or use the existing MemPalace installation:

# The toolkit uses the official mempalace library
pip install mempalace>=3.3.0

Wire a New Project

# Create .mcp.json with mempalace
cat > .mcp.json << 'EOF'
{
  "mcpServers": {
    "mempalace": {
      "type": "stdio",
      "command": "python3",
      "args": ["-m", "mempalace.mcp_server"],
      "env": {
        "PYTHONPATH": "/path/to/.mempalace/src"
      }
    }
  }
}
EOF

# Add hooks to your settings.json
# (Stop + PreCompact hooks save memory automatically)

Multi-Session Setup

All AI sessions share the same palace at /path/to/palace/:

Session	Config File	Hooks
Claude	~/.claude/settings.json	✅ Stop + PreCompact
Codex	<path-to-codex-project>/.mcp.json	✅ Via tmux session
Gemini	~/.gemini/settings.json	✅ AfterAgent + PreCompact
Qwen	~/.qwen/settings.json	✅ Stop + PreCompact + SessionEnd
claude-fcc	~/.claude-fcc/settings.json	✅ Stop + PreCompact

---

Architecture

/path/to/palace/
         │
         ├── chroma.sqlite3 (vector storage)
         ├── knowledge_graph.sqlite3 (entity relationships)
         └── wing_{project}/
             ├── hall_facts/
             ├── hall_events/
             ├── hall_discoveries/
             └── hall_preferences/

All AI sessions read/write to the same palace → unified memory across your entire ecosystem.

---

Features

Multi-Project Wiring

Each project gets a .mcp.json with mempalace configured:

# 20+ projects now wired
project-alpha/      project-beta/     project-gamma/
feature-frontend/         feature-backend/         feature-mobile/
docs-internal/        experiments sandbox/
...

Unified Memory Palace

• Claude session learns about project-alpha → Codex remembers it too
• Gemini discovers a database pattern → Qwen can search it
• No more context loss when switching between AI assistants

Template Deployment

# Template available at:
/path/to/template/.mcp.json
~/.claude/project-templates/settings.json

Copy the template to any new project — instant memory integration.

---

Why Structure Matters

Testing on real deployments:

Search Scope	Retrieval Improvement
Unfiltered search	Baseline
Within wing	+12%
Wing + hall	+24%
Wing + room	+34%

The palace structure isn't cosmetic — it's a 34% retrieval improvement.

---

Contributing

This toolkit extends the official MemPalace.

Contributions welcome for:

• Multi-project deployment patterns
• Hook automation improvements
• Team collaboration features
• Template enhancements

See CONTRIBUTING.md for guidelines.

---

License

MIT — see LICENSE.

Built on top of MemPalace by Milla Jovovich (MIT licensed).

---

Support

• Issues: GitHub Issues
• MemPalace Discord: https://discord.com/invite/ycTQQCu6kn

---

💼 Enterprise Features

Looking for team deployment, custom integrations, or support?

Services Available

Service	Description	Contact
Team Setup	Wire your entire team's AI sessions with shared memory	Email
Custom Integrations	Connect MemPalace to Slack, Jira, Notion, etc.	Email
Training & Workshops	Teach your team how to use AI memory effectively	Email
Priority Support	SLA-backed support for enterprise deployments	Email

Open Source + Commercial

• Community Edition: Free, MIT licensed (this repo)
• Enterprise Edition: Custom features, SLA, on-premise deployment

---