Comprehensive tool for creating pipelines with optional predefined steps and automatic flow instantiation.

Overview

The create_pipeline tool creates new pipeline templates with optional step definitions. Unlike other tools, this automatically creates an associated flow for immediate configuration and execution, eliminating the need for separate create_flow calls.

Parameters

• pipeline_name (string, required): Name for the new pipeline
• steps (array, optional): Array of step definitions in execution order
• flow_name (string, optional): Name for the automatically created flow (defaults to pipeline_name)
• scheduling_config (object, optional): Scheduling configuration (defaults to manual)

Step Definition Format

Each step in the steps array supports:

{
  "step_type": "fetch|ai|publish|update",
  "handler_slug": "handler_name",
  "handler_config": {
    // Handler-specific configuration
  },
  "provider": "ai_provider",     // For AI steps
  "model": "ai_model",           // For AI steps
  "system_prompt": "AI instructions" // For AI steps
}

Scheduling Configuration

Supports the same scheduling options as create_flow:

{
  "interval": "manual|hourly|daily|weekly|monthly|one_time",
  "timestamp": 1735689600  // Required for one_time
}

Usage Examples

Simple Pipeline

{
  "pipeline_name": "Basic RSS Processor"
}

Pipeline with Steps

{
  "pipeline_name": "Social Media Publisher",
  "steps": [
    {
      "step_type": "fetch",
      "handler_slug": "rss",
      "handler_config": {
        "feed_url": "https://example.com/feed.xml"
      }
    },
    {
      "step_type": "ai",
      "provider": "anthropic",
      "model": "claude-sonnet-4-20250514",
      "system_prompt": "Create engaging social media posts"
    },
    {
      "step_type": "publish",
      "handler_slug": "twitter",
      "handler_config": {
        "include_images": true
      }
    }
  ]
}

Scheduled Pipeline

{
  "pipeline_name": "Daily News Digest",
  "flow_name": "Daily Digest Flow",
  "scheduling_config": {
    "interval": "daily"
  },
  "steps": [
    {
      "step_type": "fetch",
      "handler_slug": "wordpress_api",
      "handler_config": {
        "api_url": "https://news.example.com/wp-json/wp/v2/posts"
      }
    }
  ]
}

Response

Returns comprehensive creation details:

{
  "success": true,
  "data": {
    "pipeline_id": 123,
    "pipeline_name": "Social Media Publisher",
    "flow_id": 456,
    "flow_name": "Social Media Publisher",
    "steps_created": 3,
    "flow_step_ids": [
      "pipeline_step_123_1",
      "pipeline_step_123_2",
      "pipeline_step_123_3"
    ],
    "scheduling": "manual",
    "message": "Pipeline and flow created with 3 steps. Use configure_flow_step with the flow_step_ids to set handler configurations."
  }
}

Automatic Flow Creation

This tool automatically creates a flow because:

• Pipelines are templates, flows are executable instances
• Immediate configuration is required for AI steps (provider, model, system prompt)
• Eliminates extra API calls and confusion
• Provides flow_step_ids for immediate configuration with configure_flow_step

Step Validation

Validates all step definitions:

• Checks step_type against available types
• Validates handler_slug for fetch/publish/update steps
• Ensures AI steps have required provider/model when specified
• Verifies execution order and step relationships

Integration Workflow

1. Use create_pipeline for complete pipeline + flow creation
2. Configure steps using returned flow_step_ids with configure_flow_step
3. Execute immediately with run_flow or modify scheduling with update_flow

Error Handling

Returns structured error responses for:

• Missing or invalid pipeline_name
• Malformed step definitions
• Invalid step types or handler slugs
• Scheduling configuration errors
• Permission issues

Important Notes

• Do NOT call create_flow after using this tool – a flow is already created
• Use returned flow_step_ids with configure_flow_step for handler configuration
• AI steps can include provider/model/system_prompt in the step definition
• Flow inherits pipeline name if flow_name not specified