ShedBoxAI for Data Migration

data_sources:
  old_crm:
    type: csv
    path: "old_crm_export.csv"

processing:
  format_conversion:
    old_crm:
      # Map old fields to new schema
      rename_fields:
        customer_name: "contact_name"
        phone_number: "phone"
        email_addr: "email"
      date_fields: ["created_date"]
      date_format: "%Y-%m-%d"

output:
  type: file
  path: "new_crm_import.json"
  format: json

data_sources:
  legacy_db:
    type: csv
    path: "legacy_export.csv"

processing:
  format_conversion:
    legacy_db:
      # Handle data type conversions
      number_fields: ["amount", "quantity"]

  contextual_filtering:
    legacy_db:
      - field: "status"
        condition: "!=null"
        new_name: "valid_records"

output:
  type: file
  path: "new_schema.json"
  format: json

data_sources:
  source:
    type: csv
    path: "original_data.csv"

  target:
    type: csv
    path: "migrated_data.csv"

processing:
  content_summarization:
    source:
      method: "statistical"
      fields: ["amount"]
      summarize: ["count", "sum"]

  content_summarization:
    target:
      method: "statistical"
      fields: ["amount"]
      summarize: ["count", "sum"]

ai_interface:
  model:
    type: rest
    url: "https://api.anthropic.com/v1/messages"
    method: POST
    headers:
      x-api-key: "${ANTHROPIC_API_KEY}"
      Content-Type: "application/json"
    options:
      model: "claude-sonnet-4-20250514"

  prompts:
    validate:
      system: "You are a data migration validation expert."
      user_template: |
        Validate this migration by comparing source and target:

        Source summary:
        - Records: {{ source | length }}

        Target summary:
        - Records: {{ target | length }}

        Check for: missing records, data mismatches, transformation errors.

output:
  type: file
  path: "migration_validation.md"
  format: json