Skip to main content

​
tollgate start

Starts the Tollgate proxy for a configured MCP server with full policy control.

​
Usage

dotset tollgate start [options]

​
Options

OptionDescriptionDefault
-s, --server <name>Server name from config (required)-
-c, --config <path>Path to tollgate.yaml config file./tollgate.yaml
--audit-path <path>Custom path for audit database~/.dotset/tollgate.db
--timeout <ms>Approval timeout in milliseconds60000
--dry-runEvaluate policies without executing tools-
--failure-mode <mode>Behavior when upstream failsfail-closed

​
Examples

​
Basic Usage

# Start proxy for 'postgres' server from config
dotset tollgate start -s postgres

# Use custom config path
dotset tollgate start -s postgres -c ~/configs/tollgate.yaml

​
With Options

# Start with custom timeout
dotset tollgate start -s postgres --timeout 120000

# Start with fail-readonly mode
dotset tollgate start -s filesystem --failure-mode fail-readonly

​
Dry Run Mode

# Test policies without executing tools
dotset tollgate start -s postgres --dry-run

​
Configuration File

The start command requires a tollgate.yaml configuration file. Create one with: 
dotset tollgate init

​
Example Configuration

version: "1"

defaults:
  action: prompt
  timeout: 60000

servers:
  postgres:
    command: "npx"
    args: ["-y", "@modelcontextprotocol/server-postgres"]
    env:
      DATABASE_URL: "${DATABASE_URL}"
    tools:
      "query":
        action: allow
      "execute":
        action: prompt
        analyzer: sql
      "*":
        action: deny

  filesystem:
    command: "npx"
    args: ["-y", "@anthropic/mcp-server-filesystem", "./src"]
    tools:
      "read_*":
        action: allow
      "write_*":
        action: prompt
      "delete_*":
        action: deny

​
Server Selection

The -s/--server option must match a server name defined in your config: 
servers:
  postgres:      # Use: dotset tollgate start -s postgres
    ...
  filesystem:    # Use: dotset tollgate start -s filesystem
    ...

​
When to Use

Use start when:
  • You need per-tool policies
  • Using smart content analyzers (SQL, filesystem, shell)
  • Running in production environments
  • Managing multiple MCP servers with different policies
Use wrap when:
  • Quick testing without config files
  • Uniform policy for all tools
  • One-off commands