How to Build a Full CRM Automation with .0n SWITCH Files in 10 Minutes

Here's the scenario: a new contact hits your CRM. Within the next 90 seconds, you need to enrich their data, score them against your ICP, route them to the right pipeline stage, send a personalized welcome email, and notify your sales rep on Slack.

In a traditional CRM workflow builder, this is 20+ clicks, a dozen modal dialogs, a filter condition editor that makes you want to close the laptop, and 45 minutes of your afternoon.

In 0nMCP, it's a .0n SWITCH file and one command.

Let me show you exactly how.

What Is a .0n SWITCH File?

A SWITCH file is a structured JSON document that describes an automation as a series of steps. It follows the .0n Standard — a universal config format for AI-driven workflows. Think of it as the infrastructure-as-code equivalent for business automations.

Every SWITCH file has:

• $0n — Document metadata (version, type, name)
• trigger — What kicks off the automation
• inputs — User-facing configuration variables
• launch_codes — Sensitive credentials (API keys, tokens)
• steps — The ordered list of actions to execute

Variable resolution follows a strict priority order:

{{system.}}  >  {{launch.}}  >  {{inputs.}}  >  {{step.output.}}

System variables are always trusted first. Launch codes (secrets) override input values. Step outputs are referenced by step ID.

The Automation We're Building

Here's what we want to happen when a new contact is created in the CRM:

1. Enrich — Pull company data, LinkedIn, tech stack from People Data Labs
2. Score — Check if they match our ICP (B2B, 10-500 employees, uses SaaS tools)
3. Route — Move them to "High Priority" or "Nurture" pipeline based on score
4. Email — Send a personalized welcome email using their first name and company
5. Notify — Post to the #new-leads Slack channel with a summary card

The Complete SWITCH File

{
"$0n": { "version": "1.1.0", "type": "workflow", "name": "new-contact-icp-router", "description": "Enrich, score, route, email, and notify on every new CRM contact", "created": "2026-02-27" },

"trigger": { "type": "webhook", "event": "contact.created", "source": "crm" },

"inputs": { "pipeline_id_hot": "your-hot-pipeline-id", "pipeline_id_nurture": "your-nurture-pipeline-id", "icp_min_employees": 10, "icp_max_employees": 500, "sender_name": "Mike from 0nMCP", "sender_email": "mike@rocketopp.com", "slack_channel": "#new-leads" },

"launch_codes": { "pdl_api_key": "{{env.PDL_API_KEY}}", "crm_api_key": "{{env.CRM_API_KEY}}", "slack_token": "{{env.SLACK_BOT_TOKEN}}" },

"steps": [ { "id": "enrich", "name": "Enrich Contact Data", "service": "pdl", "action": "enrich_person", "params": { "email": "{{trigger.contact.email}}", "first_name": "{{trigger.contact.first_name}}", "last_name": "{{trigger.contact.last_name}}", "company": "{{trigger.contact.company_name}}" }, "on_error": "continue" },

{ "id": "score", "name": "Score Against ICP", "service": "internal", "action": "condition", "params": { "conditions": [ { "field": "{{step.enrich.output.employee_count}}", "operator": "between", "min": "{{inputs.icp_min_employees}}", "max": "{{inputs.icp_max_employees}}" }, { "field": "{{step.enrich.output.job_company_type}}", "operator": "equals", "value": "private" } ], "logic": "AND", "output_true": "hot", "output_false": "nurture" }, "depends_on": "enrich" },

{ "id": "route", "name": "Move to Correct Pipeline", "service": "crm", "action": "create_opportunity", "params": { "contact_id": "{{trigger.contact.id}}", "pipeline_id": "{{step.score.output == 'hot' ? inputs.pipeline_id_hot : inputs.pipeline_id_nurture}}", "name": "{{trigger.contact.first_name}} {{trigger.contact.last_name}} — New Lead", "status": "open" }, "depends_on": "score" },

{ "id": "welcome_email", "name": "Send Personalized Welcome Email", "service": "crm", "action": "send_email", "params": { "contact_id": "{{trigger.contact.id}}", "from_name": "{{inputs.sender_name}}", "from_email": "{{inputs.sender_email}}", "subject": "Hey {{trigger.contact.first_name}} — welcome to the ecosystem", "body_html": "<p>Hi {{trigger.contact.first_name}},</p><p>Saw you just joined — welcome. We work with a lot of teams at {{step.enrich.output.job_company_name || trigger.contact.company_name}} size doing exactly what you're describing.</p><p>I put you in the right place. You'll hear from us shortly.</p><p>— {{inputs.sender_name}}</p>" }, "depends_on": "route" },

{ "id": "slack_notify", "name": "Notify Sales Team on Slack", "service": "slack", "action": "post_message", "params": { "channel": "{{inputs.slack_channel}}", "text": "New {{step.score.output}} lead: {{trigger.contact.first_name}} {{trigger.contact.last_name}} at {{step.enrich.output.job_company_name || trigger.contact.company_name}}\n• Employees: {{step.enrich.output.employee_count || 'unknown'}}\n• Title: {{trigger.contact.title || 'unknown'}}\n• Email: {{trigger.contact.email}}\n• Pipeline: {{step.score.output == 'hot' ? 'HIGH PRIORITY' : 'Nurture'}}", "unfurl_links": false }, "depends_on": "route" } ] } 

Breaking Down What's Happening

The Trigger Block

"trigger": {
"type": "webhook", "event": "contact.created", "source": "crm" } 

This wires the automation to the CRM's webhook system. When a contact is created, the CRM fires a POST to your 0nMCP HTTP server endpoint. The trigger payload is available as {{trigger.}} throughout all steps.

# Execute directly against a test payload
0nmcp run new-contact-icp-router.0n --input contact.email=test@example.com

Or start the HTTP server and let the CRM webhook trigger it
0nmcp serve --port 3001 

0nmcp serve --port 3001 --host 0.0.0.0

# Verify your credentials are loaded
0nmcp engine verify

Run the workflow against your real services
0nmcp run new-contact-icp-router.0n 

The engine maps your imported credentials to {{launch.}} variables automatically. No copy-pasting API keys into config files.

What Comes Next

This is one automation. But the pattern scales. Every business process you're clicking through manually is a SWITCH file waiting to be written. Once it's a file, it's:

• Versionable — git history for every change
• Portable — runs anywhere 0nMCP is installed
• Composable — combine SWITCH files into larger pipelines
• AI-editable — hand it to Claude and say "add a step that scores by company size" — it knows the schema

That last point is the one that changes everything. When your automation infrastructure is data (JSON files), AI can read it, modify it, extend it, and reason about it. When it's a drag-and-drop UI state, AI can't touch it.

Write the file. Run the file. Ship the outcome.

Get 0nMCP | .0n Standard docs | More examples