Skip to main content

CSV Column Alias Support

As of this update, AgentMap now supports multiple column name aliases for CSV files, making it more flexible when working with CSVs from different sources or following different naming conventions.

Supported Column Aliases​

The following column aliases are now supported (case-insensitive):

Required Columns​

GraphName (primary name):

  • graph_name
  • Graph
  • WorkflowName
  • workflow_name
  • workflow

Node (primary name):

  • node_name
  • NodeName
  • Step
  • StepName
  • name

Optional Columns​

AgentType (primary name):

  • agent_type
  • Agent
  • Type

Prompt (primary name):

  • prompt
  • Instructions
  • Template
  • prompt_template

Description (primary name):

  • description
  • desc
  • Details

Input_Fields (primary name):

  • input_fields
  • Inputs
  • InputFields

Output_Field (primary name):

  • output_field
  • Output
  • OutputField

Edge (primary name):

  • edge
  • next_node
  • NextNode
  • Target
  • next

Success_Next (primary name):

  • success_next
  • next_on_success
  • SuccessTarget
  • on_success

Failure_Next (primary name):

  • failure_next
  • next_on_failure
  • FailureTarget
  • on_failure

Context (primary name):

  • context
  • Config
  • Configuration

Case-Insensitive Matching​

All column names are matched case-insensitively, so these are all equivalent:

  • GraphName, graphname, GRAPHNAME, graph_name, Graph_Name

Example​

Instead of the standard format:

GraphName,Node,AgentType,Prompt,Success_Next,Failure_Next
MyWorkflow,Start,input,Enter data:,Process,Error

You can now use various aliases:

workflow_name,node_name,type,instructions,on_success,on_failure
MyWorkflow,Start,input,Enter data:,Process,Error

Or mix different styles:

graph_name,Step,agent,prompt_template,next_on_success,FailureTarget
MyWorkflow,Start,input,Enter data:,Process,Error

Implementation Details​

The column normalization is handled automatically by:

  • CSVGraphParserService - When parsing CSV files into graph specifications
  • CSVValidationService - When validating CSV file structure

The normalization happens transparently before any validation or parsing, so all downstream code continues to use the canonical column names.

Benefits​

  1. Flexibility - Work with CSVs from different sources without renaming columns
  2. Backward Compatible - Existing CSVs with standard column names continue to work
  3. User-Friendly - Use column names that make sense for your domain
  4. Case-Insensitive - No need to worry about exact capitalization

Notes​

  • Only the listed aliases are supported; other column names will be treated as unexpected
  • The canonical names (e.g., GraphName, Node) are still used internally
  • Warning messages about unexpected columns will still appear for truly unknown columns