Station

A secure, self-hosted platform for building and deploying intelligent sub-agents.

🌐 Browse Bundle Registry - Discover ready-to-use MCP bundles for Station

📚 Documentation - Complete Station documentation and guides

Lightweight runtime for deployable sub-agents that need to access internal systems, run deep in your infrastructure, and integrate seamlessly with your existing deployment processes.

Station is purpose-built for deployable sub-agents - the intelligent automation you need for infrastructure monitoring, deployment pipelines, security scanning, and day-to-day tasks that require secure access to internal systems.

Why Station Exists

When you need agents for internal work, you need more than application-focused agent platforms. You need:

  • Secure Internal Access - Agents that can safely handle database credentials, API keys, and system-level access
  • Versionable Deployment - Deployable agents that integrate with your existing deployment pipelines
  • Team-Approved Tools - Easy way to use and share the tools your team builds and approves
  • Low Footprint Runtime - Lightweight system that blends into your infrastructure without overhead

Station provides exactly this - a lightweight, secure runtime specifically designed for deployable sub-agents.

Key Features

🔒 Security by Design

  • Self-Hosted - Complete data sovereignty, no external dependencies beyond AI providers
  • Encrypted Secrets - AES encryption for credentials and sensitive configuration
  • Environment Isolation - Separate execution contexts for dev/staging/prod
  • Audit Trail - Complete tracking of agent deployments and executions

⚡ Lightweight & Integrated

  • Single 45MB Binary - No complex infrastructure or dependencies
  • SQLite + Litestream - Zero-setup database with automatic replication for production
  • GitOps Ready - Version-controlled agent configurations like infrastructure code
  • Existing Toolchain - Uses your team’s approved MCP tools and integrations

🔧 MCP Integration

  • Connects to Any MCP Client - Claude Desktop, Cursor, VS Code, and more
  • Variable Templates - Easy sharing with parameterized configurations
  • Git-Backed Registries - Install bundles directly from GitHub repositories
  • Add Any MCP Tools - Assign filesystem, AWS, database, or custom tools to agents

Quick Start

Get production-ready security agents in 3 simple steps:

# 1. Install Station
curl -fsSL https://raw.githubusercontent.com/cloudshipai/station/main/install.sh | bash
stn init --ship

# 2. Install DevOps Security Bundle
stn bundle install https://github.com/cloudshipai/registry/releases/latest/download/devops-security-bundle.tar.gz security

# 3. Sync and connect to Claude Code
stn sync
echo '{"mcpServers":{"station":{"command":"stn","args":["stdio"]}}}' > ~/.claude_desktop_config.json

✅ Ready! Open Station UI and start using security agents through Claude Code.

Station Interface

Architecture Overview

Station provides multiple interfaces for agent management:

  • CLI - Command-line interface for development and scripting
  • REST API - HTTP API for integration with existing systems
  • SSH/TUI - Terminal-based admin interface
  • MCP Server - Model Context Protocol for AI client integration
graph TB
    %%{init: {"theme": "default", "themeVariables": {"fontSize": "14px"}, "flowchart": {"curve": "basis", "padding": 20}}}%%
    subgraph "Station Runtime"
        CLI["stn CLI Commands"]
        SSH["SSH Terminal UI"]
        HTTP["HTTP API Server"]
        MCP["MCP Server"]
    end
    
    subgraph "Core Services"
        Scheduler["SchedulerService"]
        ExecutionQueue["ExecutionQueueService"]
        Webhook["WebhookService"]
        Agent["AgentService"]
        ToolDiscovery["ToolDiscoveryService"]
    end
    
    subgraph "Data Layer"
        Repository["Repository Layer"]
        Config["Config Files"]
        SQLite["SQLite Database"]
    end
    
    subgraph "AI & Integration"
        GenKit["GenKit Framework"]
        ExternalMCP["External MCP Servers"]
        Gemini["Gemini Provider"]
        OpenAI["OpenAI Provider"]
    end
    
    CLI --> Agent
    SSH --> Agent
    HTTP --> Agent
    MCP --> Agent
    
    Agent --> ExecutionQueue
    Agent --> ToolDiscovery
    ExecutionQueue --> Scheduler
    Scheduler --> Webhook
    
    Agent --> Repository
    Repository --> Config
    Repository --> SQLite
    
    Agent --> GenKit
    ToolDiscovery --> ExternalMCP
    GenKit --> Gemini
    GenKit --> OpenAI

All interfaces connect to the core agent management system with MCP tool integration, configuration management, and security services.

Next Steps