CyberMem
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@CyberMemremember that my favorite color is blue"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Features
Feature | Description |
Model Context Protocol | Native Model Context Protocol support for Claude, Cursor, and other AI clients |
Multi-Platform | Deploy on Mac, Raspberry Pi, or Cloud VPS with one command |
Infrastructure as Code | Production-ready Ansible Playbooks, Helm Charts, Docker Compose |
Observability | Built-in SQLite activity metrics, beautiful time-series charts, audit logs |
Ansible-First Prod | Senior DevOps Workflow: Automated deployment, health-checks, and state management |
Hybrid CI/CD | Self-Hosted RPi Runner for native 64-bit ARM builds + GitHub Cloud for x86 |
Security | Traefik reverse proxy, Tailscale Funnel for zero-config HTTPS |
To install CyberMem on your local machine, run:
npx @cybermem/cli installand follow the instructions in terminal.
Full Quick Start guide for every platform is available at cybermem.dev/#quickstart.
Why CyberMem?
Problem: Your AI tools (Claude, Cursor, Antigravity) don't share memory. Each session starts fresh.
Solution: CyberMem gives them a shared, persistent memory layer.
Without CyberMem | With CyberMem |
Claude forgets your project context | All tools remember your preferences |
Cursor doesn't know your coding style | Context persists across sessions |
Each tool has separate knowledge | One unified memory for all AI agents |
For Platform Engineers: CyberMem demonstrates advanced Infrastructure practices:
IaC Automation: CLI generates Docker Compose, Ansible, or Helm depending on the target.
Hybrid CI: Leverages a private Raspberry Pi runner to bypass QEMU overhead, achieving native ARM64 build speeds.
Zero-Trust Access: Integrates Tailscale Funnel for secure, public access without port forwarding.
For MCP Engineers & Hiring Managers
CyberMem is a reference implementation of production MCP patterns:
π Multi-tenant auth: Traefik ForwardAuth extracts MCP client identity into audit logs
π Zero-overhead observability: Vector pipeline converts logs β Prometheus metrics
π Multi-platform IaC: CLI auto-generates Docker Compose / Ansible / Helm per target
βΈοΈ Production deployment: Kubernetes with rolling updates, health checks, PVC
π§ Hybrid CI/CD: Self-hosted ARM64 runner bypasses QEMU 10x slowdown
551+ commits β’ 25 releases β’ Full E2E testing
If you're building MCP infrastructure at companies like Descope, Anthropic, or similar β use this as a production pattern library.
π§ Open to MCP/Agentic Infrastructure roles: mikhailkogan17@gmail.com
Architecture Overview
---
config:
theme: base
wrappingWidth: 290
fontSize: 11
---
graph TD
subgraph CI["π Hybrid CI/CD Pipeline"]
GHA["**GitHub Actions**<br/>(Control Plane)"]
Cloud["**GitHub Cloud**<br/>(AMD64 Builds)"]
RPiR["**Self-Hosted RPi**<br/>(Native ARM64 Builds)"]
GHA --> Cloud
GHA --> RPiR
end
subgraph Gen["ποΈ IaC Engine"]
CLI["**CLI**"]
Templates["**Infrastructure Templates**"]
CLI --> Templates
Compose["**Docker Compose**<br/>(Local)"]
Ansible["**Ansible**<br/>(IoT/Edge)"]
Helm["**Helm Charts**<br/>(Cloud/K8s)"]
Templates --> Compose
Templates --> Ansible
Templates --> Helm
end
subgraph Runtime["βοΈ CyberMem Runtime"]
Traefik["**Traefik**<br/>(Reverse Proxy)"]
LE["**Log Exporter**<br/>(Access Logs)"]
DBE["**DB Exporter**<br/>(SQLite Metrics)"]
Dash["**Dashboard**<br/>(Monitoring UI)"]
OM["**Core API**<br/>(Memory Engine)"]
DB["**SQLite / Postgres**<br/>(Persistence)"]
Traefik -->|Logs| LE
Traefik -->|API| OM
OM --> DB
LE --> DB
DB --> DBE
DBE --> Dash
end
Cloud -.-> Runtime
RPiR -.-> Runtime
Compose -.-> Traefik
Ansible -.-> Traefik
Helm -.-> TraefikCLI Reference
CyberMem CLI provides a standardized set of commands for complete lifecycle management:
npx @cybermem/cli install # Install/Initialize services (Mac/RPi/VPS)
npx @cybermem/cli uninstall # Stop and teardown services
npx @cybermem/cli upgrade # Pull latest images and update instance
npx @cybermem/cli backup # Create a data backup (.tar.gz)
npx @cybermem/cli restore # Restore from a backup file
npx @cybermem/cli reset # Wipe database (Destructive!)
npx @cybermem/cli dashboard # Open monitoring dashboardAnsible-First Prod: For Raspberry Pi or remote servers, the CLI automatically leverages Ansible to ensure state-consistent, safe, and verifiable deployments.
Project Structure (Monorepo)
cybermem/
βββ packages/
β βββ cli/ # Command-line tool (TypeScript)
β β βββ src/ # CLI logic
β β βββ templates/ # β Infrastructure templates
β βββ mcp/ # MCP Server & Core Engine (TypeScript)
β β βββ src/ # Tooling & Memory Logic
β βββ dashboard/ # Monitoring UI (Next.js)
βββ docs/ # Documentation sources
βββ .github/
β βββ workflows/ # β CI/CD pipelines
βββ README.mdKey innovation: packages/cli/templates/ contains the infrastructure-as-code templates.
The CLI reads these, interpolates variables, and generates production configs.
Documentation
Full documentation available at docs.cybermem.dev:
Guide | Description |
Mac/Linux development environment | |
Production standard for RPi/Edge | |
Edge deployment with Tailscale | |
Production Kubernetes deployment | |
Connect Claude, Cursor, and more |
π οΈ Architecture Narratives
Why Traefik for ForwardAuth?
Standard Node.js auth middlewares often fail on underpowered Edge devices (RPi) or cause high latency. CyberMem uses Traefik as a Reverse Proxy to handle authentication at the networking layer. This allows the Core API to remain "clean" and deterministic, while Traefik extracts identity headers (X-Client-Name) into audit logs before the request even hits the application.
Why Ansible for RPi but Helm for Cloud?
We follow the Infrastructure Appropriateness principle.
RPi/Edge: Needs mutable state management and OS-level hardening (docker-compose, systemd). Ansible ensures idempotent state without the overhead of a control plane.
Cloud/VPS: Scaling and high availability are paramount. Helm allows us to leverage Kubernetes native primitives (Ingress, PVC, HPA) for a truly elastic platform.
Verification & Proof-of-Work
We use tools/test-k8s.sh and the CyberMem Gatekeeper to guarantee that every release is stable. Below is the raw console verification of a production-grade deployment.
1. Kubernetes Resource Tree (Architecture Proof)
NAMESPACE: cybermem
NAME READY STATUS RESTARTS AGE
pod/cybermem-dashboard-6dd67f5586-djwwh 1/1 Running 0 2m
pod/cybermem-openmemory-65fdf6d85c-g628g 1/1 Running 0 2m
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
cybermem-lb LoadBalancer 10.43.0.1 127.0.0.1 8626/TCP 2m
cybermem-dashboard ClusterIP 10.43.79.62 <none> 3000/TCP 2m
cybermem-openmemory ClusterIP 10.43.95.212 <none> 8080/TCP 2m2. Ansible Idempotency (Operational Maturity)
# Proof of zero-drift state management on Raspberry Pi
ansible-playbook -i inventory/hosts.ini playbooks/deploy-cybermem.yml
PLAY [Deploy CyberMem to Raspberry Pi] **
TASK [Gathering Facts] **
ok: [raspberrypi.local]
TASK [cybermem : Pull latest images from GHCR] **
ok: [raspberrypi.local] => (changed=false)
TASK [cybermem : Start services] **
ok: [raspberrypi.local] => (changed=false)
PLAY RECAP **
raspberrypi.local : ok=15 changed=0 unreachable=0 failed=0 skipped=0Contributing
Contributions are welcome! See CONTRIBUTING.md for development setup and guidelines.
License
MIT
Created by Mikhail Kogan
π§ Open to MCP/Agentic Infrastructure roles: mikhailkogan17@gmail.com
π LinkedIn β’ CV (PDF)
This server cannot be installed
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/mikhailkogan17/cybermem'
If you have feedback or need assistance with the MCP directory API, please join our Discord server