/station

name: station
description: Use Station CLI (`stn`) for AI agent orchestration - creating agents, running tasks, managing environments, and deploying agent teams. Prefer CLI for file operations and exploration; use MCP tools for programmatic agent execution and detailed queries.

Station CLI

Station is a self-hosted AI agent orchestration platform. You interact with it via the `stn` CLI or MCP tools (41+ available via `stn stdio`).

When to Use CLI vs MCP Tools

| Task | Use CLI | Use MCP Tool | |------|---------|--------------| | Create/edit agent files | `stn agent create`, edit `.prompt` files | - | | Run an agent | `stn agent run <name> "<task>"` | `call_agent` | | List agents/environments | `stn agent list`, `stn env list` | `list_agents`, `list_environments` | | Add MCP servers | `stn mcp add <name>` | `add_mcp_server_to_environment` | | Sync configurations | `stn sync <env>` | - | | Install bundles | `stn bundle install <url>` | - | | Inspect runs | `stn runs list` | `inspect_run`, `list_runs` | | Deploy | `stn deploy <env>` | - | | Start services | `stn serve`, `stn jaeger up` | - |

**Rule of thumb**: CLI for setup, file operations, deployment. MCP tools for programmatic execution and queries within conversations.

Quick Reference

Initialization

# Initialize Station with AI provider
stn init --provider openai --ship       # OpenAI with Ship filesystem tools
stn init --provider anthropic --ship    # Anthropic (requires OAuth: stn auth anthropic login)
stn init --provider gemini --ship       # Google Gemini

# Initialize in specific directory (git-backed workspace)
stn init --provider openai --config ./my-workspace

# Start Jaeger for observability
stn jaeger up                           # View traces at http://localhost:16686

Agent Management

# List agents
stn agent list                          # All agents in default environment
stn agent list --env production         # Agents in specific environment

# Show agent details
stn agent show <agent-name>             # Full configuration

# Run an agent
stn agent run <name> "<task>"           # Execute with task
stn agent run incident-coordinator "High latency on API"
stn agent run cost-analyzer "Analyze this week's AWS spend" --env production
stn agent run my-agent "task" --tail    # Follow output in real-time

# Delete agent
stn agent delete <name>

Environment Management

# List environments
stn env list

# Sync file configurations to database
stn sync default                        # Sync default environment
stn sync default --browser              # Secure input for secrets (recommended for AI)
stn sync default --dry-run              # Preview changes
stn sync default --validate             # Validate only

MCP Server Configuration

# Add MCP server
stn mcp add <name> --command <cmd> --args "<args>"

# Examples
stn mcp add filesystem --command npx --args "-y,@modelcontextprotocol/server-filesystem,/path"
stn mcp add github --command npx --args "-y,@modelcontextprotocol/server-github" --env "GITHUB_TOKEN={{.TOKEN}}"
stn mcp add playwright --command npx --args "-y,@playwright/mcp@latest"

# Add OpenAPI spec as MCP server
stn mcp add-openapi petstore --url https://petstore3.swagger.io/api/v3/openapi.json

# List and manage
stn mcp list                            # List configurations
stn mcp tools                           # List available tools
stn mcp status                          # Show sync status
stn mcp delete <config-id>              # Remove configuration

Bundle Management

# Install bundle from URL or CloudShip
stn bundle install <url-or-id> <environment>
stn bundle install https://example.com/bundle.tar.gz my-env
stn bundle install devops-security-bundle security

# Create bundle from environment
stn bundle create <environment>
stn bundle create default --output ./my-bundle.tar.gz

# Share bundle to CloudShip
stn bundle share <environment>

# Export required variables from bundle (for CI/CD)
stn bundle export-vars ./my-bundle.tar.gz --format yaml
stn bundle export-vars ./my-bundle.tar.gz --format env
stn bundle export-vars <cloudship-bundle-id> --format yaml

Workflow Management

# List workflows
stn workflow list
stn workflow list --env production

# Run workflow
stn workflow run <name>
stn workflow run incident-response --input '{"severity": "high"}'

# Manage approvals (for human-in-the-loop)
stn workflow approvals list
stn workflow approvals approve <approval-id>
stn workflow approvals reject <approval-id> --reason "Not authorized"

# Inspect and validate
stn workflow inspect <run-id>
stn workflow validate <name>
stn workflow export <name> --output workflow.yaml

Server & Deployment

# Start Station server (web UI at :8585)
stn serve
stn serve --dev                         # Development mode

# Docker container mode
stn up                                  # Interactive setup
stn up --bundle <bundle-id>             # Run specific bundle
stn status                              # Check container status
stn logs -f                             # Follow logs
stn down                                # Stop container

# DEPLOY TO CLOUD (3 methods)
# Method 1: Local environment
stn deploy <environment> --target fly   # Deploy to Fly.io
stn deploy production --target k8s      # Deploy to Kubernetes
stn deploy production --target ansible  # Deploy via Ansible (SSH + Docker)

# Method 2: CloudShip bundle ID (no local environment needed)
stn deploy --bundle-id <uuid> --target fly
stn deploy --bundle-id <uuid> --target k8s --name my-station

# Method 3: Local bundle file
stn deploy --bundle ./my-bundle.tar.gz --target fly
stn deploy --bundle ./my-bundle.tar.gz --target k8s

# Deploy flags
--target        fly, kubernetes/k8s, ansible (default: fly)
--bundle-id     CloudShip bundle UUID (uses base image)
--bundle        Local .tar.gz bundle file
--name          Custom app name
--region        Deployment reg