v0.1.10 is out!

Local AI workspaces
built for efficiency.

Isolated AI workspaces that remember, think, and run — on your machine.

$ npm install -g memoryblock
mblk status
mblk status
home (active)
🕒 Last pulse: 2 mins ago
🧠 Memory: 4.2k tokens

ops-monitor (active)
🕒 Last pulse: 12 mins ago
🧠 Memory: 1.8k tokens

Total tokens used today: 12,480

Built for Performance

Memoryblock reimagines AI agents as localized, persistent workers.

🧊

Block Architecture

Isolated workspaces with dedicated memory, identity, and tools. Copy a folder to move an entire monitor setup.

Lightweight

No heavy runtimes or Docker required. Runs instantly on Bun or Node.js. High efficiency, low latency.

🧠

Persistent Memory

Monitors remember context across sessions. Smart summarization keeps prompts lean while maintaining deep knowledge.

📡

Multi-Channel

Conversations sync seamlessly across CLI, Native Web Dashboard, and Chat platforms.

🛡️

Safe by Default

Sandboxed execution with human approval in loop for critical commands. Unique permissions across blocks.

🔌

Model Agnostic

Native adapters for OpenAI, Anthropic, Gemini, and Bedrock. Switch models per task with a single config line.

Autonomic Pulse System

Background workers that wake up, execute tasks, and alert you only when needed. Scripts, crons, alerts, and webhooks running silently without wasting human attention.

/* Monitor task: Registering autonomic pulse */
set_pulse({
  id: "uptime-ping",
  type: "cron",
  cron_expression: "0 * * * *",
  instruction: "Verify server connection"
});

Ready to deploy?

Ship your first isolated agent in less than 60 seconds.