Skip to content
vviinn DocsDocs Beta

Relevant Columns

list_relevant_columns is a discovery tool that tells the agent which product columns are relevant for a given site. This tool helps understand which product attributes can be used for filtering, building facets, or displaying information to the shopper.

Relevance for a column is derived from the union of enabled_dynamic_filters across all four search campaign types (Text Search, Visual Search, Similar Products, and Cross-selling). Each column returned is guaranteed to be present in the site’s product schema.

The label for each column is determined by a majority vote among the titles configured for that column across all contributing campaigns. Ties are broken by campaign type precedence (e.g., text search takes priority over visual search) and then lexicographically by the label string.

Once a relevant column’s name is known, the describe_column tool can be used to retrieve its catalog-wide values, such as enumerated options, numeric ranges, or category hierarchies.

list_relevant_columns takes no arguments.

{
"title": "RelevantColumnsRequest",
"type": "object"
}

The response returns a list of columns, where each McpRelevantColumn specifies the name, type, and an optional human-readable label.

{
"title": "RelevantColumnsResponse",
"type": "object",
"properties": {
"columns": {
"type": "array",
"items": {
"$ref": "#/$defs/McpRelevantColumn"
}
}
},
"required": [
"columns"
],
"$defs": {
"McpRelevantColumn": {
"type": "object",
"properties": {
"name": {
"type": "string",
"description": "The column's name — reference it to act on the column, e.g. as a static filter."
},
"type": {
"type": "string",
"description": "The column's value type, e.g. `Name`, `Tree` (category paths), `Enumerated`, `Float`."
},
"label": {
"type": [
"string",
"null"
],
"description": "Human-readable label for the column. Omitted when none is configured."
}
},
"required": [
"name",
"type"
]
}
}
}