Claude Desktop Usage Patterns¶

You: "Create a webhook workflow in staging that:
- Receives POST requests at /customer-signup
- Validates email and name fields are present
- Sends welcome email using Gmail credentials
- Stores customer data in PostgreSQL customers table
- Returns success/error JSON response"

Claude: I'll create a comprehensive customer signup workflow in staging with validation, email, and database storage.

[MCP tool call: create_workflow]
{
  "instance": "staging",
  "name": "Customer Signup Handler",
  "nodes": [
    {
      "name": "Webhook",
      "type": "n8n-nodes-base.webhook",
      "parameters": { "path": "customer-signup" }
    },
    {
      "name": "Validate Fields",
      "type": "n8n-nodes-base.if",
      "parameters": {
        "conditions": {
          "boolean": [
            { "value1": "={{ $json.email }}", "operation": "isNotEmpty" },
            { "value1": "={{ $json.name }}", "operation": "isNotEmpty" }
          ]
        }
      }
    },
    {
      "name": "Send Welcome Email",
      "type": "n8n-nodes-base.gmail"
    },
    {
      "name": "Store in Database",
      "type": "n8n-nodes-base.postgres"
    }
  ]
}

✅ Workflow created successfully in staging

Details:
- ID: 789
- Webhook URL: https://staging.../webhook/customer-signup
- Nodes: 4 (Webhook, Validation, Gmail, PostgreSQL)
- Status: Inactive

Would you like me to activate it or help you test it first?

You: "Make a workflow for signups"

Claude: I can help create a signup workflow. To provide the best solution, I need more information:

1. Which instance? (production, staging, development)
2. How should it be triggered? (webhook, schedule, manual)
3. What data should it process?
4. Where should the data go? (database, email, API)
5. What integrations are needed?

Could you provide these details?

Step 1: Create Foundation
You: "Create a basic webhook workflow in staging at /order-webhook"

Claude: [Creates webhook workflow with basic structure]
✅ Created webhook workflow (ID: 123)

Step 2: Add Processing
You: "Add a node that extracts order_id, amount, and customer_email from the webhook data"

Claude: [Adds code node for data extraction]
✅ Added data extraction node

Step 3: Add Validation
You: "Add validation to check that amount is greater than 0 and customer_email is valid"

Claude: [Adds validation node]
✅ Added validation with error handling

Step 4: Add Integration
You: "When validation passes, send a Slack notification to #orders channel"

Claude: [Adds Slack node]
✅ Added Slack integration

Step 5: Test
You: "Show me the workflow structure and suggest how to test it"

Claude: [Displays workflow and provides test curl command]

You: "Workflow 456 in production is failing with errors. Help me debug it."

Claude: I'll help debug workflow 456. Let me investigate.

[Step 1: Check Recent Executions]
[MCP tool call: list_executions]
{
  "instance": "production",
  "workflowId": "456",
  "finished": false,
  "limit": 5
}

Found 3 failed executions in the past hour.

[Step 2: Get Failure Details]
[MCP tool call: get_execution]
{
  "instance": "production",
  "id": "9876",
  "includeData": true
}

**Diagnosis:**

Issue Found:
- Error: "ECONNREFUSED"
- Node: "Fetch Customer Data"
- URL: http://localhost:3000/api/customers
- Line: HTTP Request node, line 45

Root Cause:
The workflow is trying to connect to localhost, but your n8n instance is hosted in the cloud and cannot reach local development servers.

**Solutions:**

1. **Update to Production API** (Recommended)
   Change URL to: https://api.yourcompany.com/customers

2. **Use ngrok for Testing**
   Expose local server: ngrok http 3000
   Update URL to ngrok URL

3. **Switch to Staging**
   Test against staging API instead

Would you like me to update the URL to your production API?

Stage 1: Development
You: "Create a customer onboarding workflow in development"

Claude: [Creates in development instance]
✅ Created in development (ID: 101)

Stage 2: Testing in Development
You: "Activate workflow 101 in development and show me how to test it"

Claude: [Activates and provides test instructions]

Stage 3: Promote to Staging
You: "The workflow works in development. Create the same workflow in staging."

Claude: I'll replicate the workflow to staging for QA testing.

[Gets workflow from development]
[Creates identical workflow in staging]

✅ Workflow replicated to staging (ID: 201)
Note: Credentials and environment-specific settings may need adjustment.

Stage 4: QA in Staging
You: "List all executions for workflow 201 in staging from today"

Claude: [Shows staging execution history]

Stage 5: Deploy to Production
You: "Staging tests passed! Create this workflow in production but keep it inactive until I review it."

Claude: [Creates in production, keeps inactive]
✅ Created in production (ID: 301) - Status: Inactive

Would you like me to:
1. Activate it now
2. Show configuration for final review
3. Set up monitoring

Development (Rapid Testing):
You: "Create a quick test workflow in development with manual trigger"
Purpose: Fast iteration, experimentation

Staging (Pre-Production QA):
You: "Deploy the customer-onboarding workflow to staging for QA testing"
Purpose: Integration testing, validation

Production (Live System):
You: "Carefully create the payment-notification workflow in production. Keep it inactive until I've reviewed the configuration and tested it."
Purpose: Controlled deployment, safety

Scenario: Cleanup Old Workflows

You: "List all inactive workflows in production"

Claude: [MCP tool call: list_workflows]
{
  "instance": "production",
  "active": false
}

Found 5 inactive workflows in production:
1. Old Email Campaign (ID: 101) - Last updated: 6 months ago
2. Deprecated Slack Bot (ID: 102) - Last updated: 4 months ago
3. Test Workflow (ID: 103) - Last updated: 1 month ago
4. Legacy Data Sync (ID: 104) - Last updated: 5 months ago
5. Unused Webhook (ID: 105) - Last updated: 2 months ago

You: "Delete workflows 103, 104, and 105 - they're no longer needed"

Claude: I'll delete the three unused workflows from production.

[Deletes workflows sequentially]

✅ Deleted 3 workflows:
- Test Workflow (103) ✓
- Legacy Data Sync (104) ✓
- Unused Webhook (105) ✓

2 inactive workflows remain:
- Old Email Campaign (101)
- Deprecated Slack Bot (102)

Would you like me to review these before deletion?

By Name:
You: "Find all workflows in production that have 'customer' in the name"

Claude: [Searches and lists matching workflows]

By Tag:
You: "Show me workflows tagged 'email-automation' in staging"

Claude: [Lists workflows with specific tag]

By Status:
You: "List active workflows in production that haven't executed in the last 7 days"

Claude: I'll check active workflows and their execution history.

[Lists workflows]
[For each, checks last execution]

Found 3 active workflows with no recent executions:
1. Workflow 201 - Last run: 12 days ago
2. Workflow 305 - Last run: 9 days ago
3. Workflow 412 - Last run: 15 days ago

These workflows are active but not triggering. Possible reasons:
- Schedule not matching expectations
- Webhook not receiving requests
- Trigger condition not being met

Would you like me to investigate each one?

You: "Get workflow 789 from production, analyze its structure, then help me add comprehensive error handling"

Claude: I'll retrieve and analyze workflow 789.

[MCP tool call: get_workflow]
{
  "instance": "production",
  "id": "789"
}

**Current Workflow Structure:**

Name: "Order Processing"
Nodes: 4
Flow:
1. Webhook → Receives orders
2. HTTP Request → Validates with payment API
3. PostgreSQL → Stores order
4. Email → Sends confirmation

**Analysis:**

Missing Components:
✗ No error handling for API failures
✗ No retry logic
✗ No failure notifications
✗ No data validation

**Recommended Improvements:**

1. Add Error Trigger node to catch failures
2. Add retry logic (3 attempts with exponential backoff)
3. Send Slack alert on persistent failures
4. Add data validation before processing
5. Log all errors to monitoring system

Would you like me to implement these improvements?

You: "Yes, add all recommended error handling"

Claude: I'll update the workflow with comprehensive error handling.

[Updates workflow with new nodes and error handling]

✅ Workflow updated with:
- Error Trigger node for API failures
- 3x retry logic with exponential backoff
- Slack notifications to #ops-alerts
- Input validation node
- Error logging to monitoring system

The workflow is now production-ready with robust error handling.

You: "Fix my workflow"

You: "Workflow 123 in production returns 'Invalid JSON' errors when processing webhook data. Can you help debug it?"

You: "Create a workflow"

You: "Create a workflow in staging for testing before production deployment"

You: "Create a Slack workflow"

You: "I've configured Slack credentials in n8n as 'Slack Production'. Create a workflow that uses these credentials to send notifications."

You: "Update that webhook we talked about yesterday"

You: "Update workflow 789 (Customer Signup Webhook) in staging to add email validation before processing"

You: "Create in production and activate immediately"

You: "Create in staging first, let me test it thoroughly, then I'll ask you to deploy to production"

"Create webhook at /api/orders in staging that validates order data and stores in PostgreSQL"
"List all workflows tagged 'customer-service' in production"
"Show executions for workflow 123 from the last 24 hours that failed"

"Make something"
"Show me workflows"
"Check that thing"

"I'm testing this in staging before promoting to production"
"This workflow processes Stripe webhooks and needs to handle payment events"
"The webhook receives GitHub push notifications and triggers deployments"

"Update workflow 456 in production"
"Get execution details for execution 9876"
"Delete workflow 789 from staging"

"Update that customer workflow"
"Check the last thing that ran"
"Delete the old one"

"Delete workflow 789 from production - I've confirmed it's no longer needed and backed up the configuration"
"Deactivate these workflows after you list them: [IDs]"

"Delete everything old"
"Remove all test workflows"

Step 1: "Create basic webhook structure"
Step 2: "Add data validation"
Step 3: "Add Slack notification on errors"
Step 4: "Test and show me the results"

Single paragraph with 10 different requirements mixed together

You: "I need to create a workflow using Slack, but I haven't configured Slack credentials yet"

Claude: I cannot create credentials directly (n8n blocks this for security), but I'll guide you through the setup:

**Slack Credential Setup:**

1. **Create Slack App:**
   - Go to: https://api.slack.com/apps
   - Click "Create New App"
   - Choose "From scratch"
   - Name: "n8n Integration"
   - Select your workspace

2. **Configure Permissions:**
   - OAuth & Permissions → Add Bot Token Scopes:
     - chat:write (send messages)
     - chat:write.public (post to public channels)
     - files:write (upload files)

3. **Install to Workspace:**
   - Click "Install to Workspace"
   - Authorize
   - Copy Bot User OAuth Token (starts with xoxb-)

4. **Add to n8n:**
   - n8n UI → Settings → Credentials
   - Add Credential → "Slack OAuth2 API"
   - Paste token
   - Save as "Slack Production"

Once configured, let me know and I'll create the Slack workflow.

You: "Done! Credentials saved as 'Slack Production'"

Claude: Perfect! I'll create the Slack notification workflow now.
[Creates workflow using Slack credentials]

You: "Workflow 456 in production takes 2 minutes to complete. Can you analyze and optimize it?"

Claude: I'll analyze workflow performance.

[Gets workflow structure and execution data]

**Performance Analysis:**

Workflow: "Daily Data Sync"
Average Duration: 125 seconds
Nodes: 6

**Bottlenecks Identified:**

1. **HTTP Request Node** (90 seconds)
   - Fetching 10,000 records individually
   - Solution: Use batch API endpoint
   - Expected savings: ~75 seconds

2. **Loop Processing** (30 seconds)
   - Processing records sequentially
   - Solution: Use Split In Batches for parallel processing
   - Expected savings: ~20 seconds

3. **Multiple Database Writes** (5 seconds)
   - Individual INSERT statements
   - Solution: Use bulk INSERT
   - Expected savings: ~3 seconds

**Total Expected Improvement:**
125s → ~25s (80% faster)

**Recommended Actions:**
1. Replace individual HTTP requests with bulk API call
2. Add Split In Batches node for parallel processing
3. Implement bulk database operations

Would you like me to implement these optimizations?