AgentOS — Operating System for AI Agents

The infrastructure layer that gives AI agents persistent memory, task management, and team collaboration capabilities. Think of it as the SQLite of the AI agent world — one MCP server, zero config, every service your agent needs.

The Problem We Solve

AI agents today have a critical limitation: they forget everything when the conversation ends. Without persistent infrastructure, agents cannot:

• Remember what they did yesterday

• Track complex, multi-step tasks across sessions

• Collaborate with other agents or human teams

• Maintain audit trails for compliance

• Prevent runaway loops and resource exhaustion

AgentOS fixes this by providing the missing operating system layer for AI agents.

What AgentOS Does

AgentOS is an MCP (Model Context Protocol) server that gives AI agents the same infrastructure that traditional software takes for granted: persistent storage, state management, collaboration tools, and monitoring.

Core Capabilities

1. Persistent Memory Your agents can store and retrieve data across sessions. Set a value today, retrieve it next week. Namespaces keep data organized, and TTL support enables temporary storage that auto-expires.

2. Task Management Break complex workflows into trackable tasks with checkpoints. Create a task, save progress at key milestones, rollback if something goes wrong, and maintain a complete history of what happened.

3. Resource Protection Set budgets to prevent runaway loops and resource exhaustion. Automatic loop detection catches infinite cycles before they drain your compute budget.

4. Team Collaboration Create isolated workspaces for teams. Manage members with role-based permissions (admin, manager, member, viewer). Share resources while maintaining security boundaries.

5. Enterprise Security Complete audit trails for SOC2, GDPR, and HIPAA compliance. Data encryption at rest. Role-based access control with granular permissions.

6. Operations & Monitoring Real-time performance metrics, intelligent alerting, capacity planning, and business intelligence reporting. Know when to scale before you hit limits.

7. Workflow Automation Template engine for reusable document and workflow patterns. Convert between formats (Markdown, HTML, CSV, JSON, YAML, XML). Structured logging with search and export.

What You Can Build

With AgentOS, your AI agents can now:

• Remember project context across multiple sessions over days or weeks

• Track complex deployments with rollback capability if something fails

• Collaborate as a team with shared workspaces and role-based access

• Maintain compliance with complete audit trails for every action

• Prevent disasters with automatic loop detection and budget enforcement

• Scale confidently with monitoring, alerting, and capacity planning

Quick Start

Add to your Claude Desktop or any MCP client:

{
  "mcpServers": {
    "agentos": {
      "command": "npx",
      "args": ["-y", "agentos-mcp"]
    }
  }
}

No configuration needed. AgentOS automatically creates a local SQLite database at ~/.agentos/agentos.db for persistent storage.

Example Use Cases

Personal Productivity

• Store research notes that persist across conversations

• Track long-running projects with task checkpoints

• Log all activities for later reference and analysis

Software Development

• Manage multi-step deployment workflows with rollback capability

• Track feature development across multiple coding sessions

• Maintain audit logs for compliance requirements

Team Collaboration

• Share knowledge bases across team members

• Coordinate complex projects with shared task tracking

• Maintain consistent templates and workflows

Enterprise Operations

• Ensure compliance with complete audit trails

• Monitor system health and performance metrics

• Plan capacity based on usage patterns and predictions

Architecture

• Protocol: MCP (Model Context Protocol) — works with Claude, Cursor, and any MCP-compatible client

• Storage: SQLite with sql.js — portable, zero-config, no external dependencies

• Security: AES-256-GCM encryption, role-based access control, audit logging

• Deployment: npm install, Docker, or self-hosted — your choice

Installation

# Install globally
npm install -g agentos-mcp

# Run standalone
agentos-mcp

# Or use with npx (no install needed)
npx -y agentos-mcp

Development

# Clone and install
git clone https://github.com/netflypsb/agentos.git
cd agentos
npm install

# Development mode
npm run dev

# Build
npm run build

# Test
npm test

Works With

Claude Desktop · Claude Code · Cursor · Windsurf · VS Code · Zed · any MCP client

Support

• Documentation: GitHub Repository

• Issues: GitHub Issues

• MCP Marketplace: Available on the MCP Marketplace for enhanced features

License

MIT License — see LICENSE file for details.