Nexadata MCP Server Tool Reference

list_orgs lists every organization you are a member of. The currently preferred organization is marked, and a single-org user has that organization selected automatically.

select_org sets your preferred organization for subsequent requests and confirms the organization name, ID, and slug.

list_workspaces lists all workspaces in an organization, including each workspace's environments and lifecycle status.

list_connections lists data connections, with optional filters for name and connection type such as sftp or s3. Results are limited to 20.

search_datasets searches for datasets in an organization, with optional name and source type filters. Results are ordered by most recently updated and limited to 50.

preview_dataset downloads and displays the first 10 rows of a dataset, along with metadata and total row count.

list_workflows lists workflows within a specific workspace, with an optional name filter. Results are limited to 50.

workflow_lookup finds workflows by name across an organization and lists each workflow's pipelines, grouped into output pipelines and supporting pipelines. The primary output pipeline is tagged.

create_workflow creates a new workflow in a workspace and environment.

add_dataset_to_workflow associates a dataset with a workflow so it can feed pipelines.

create_output_pipeline creates an output pipeline within a workflow. It takes either a dataset or a supporting pipeline as its input. This is the tool to start with when building a pipeline unless you specifically need to pre-process data first.

create_supporting_pipeline creates a supporting pipeline within a workflow. Supporting pipelines process data from a dataset or another supporting pipeline, and their output can feed output pipelines. Use one when you need to pre-process data before joining it into an output pipeline.

create_default_output creates a default "Managed by Nexadata" output for a workflow that has no outputs defined. Because a workflow needs a default output before it can run, this is a quick way to make a workflow executable.

get_pipeline_transformations returns every transformation in a pipeline with full detail, including inputs, columns, data sources, and mapping groups. Use it to understand the complete transformation chain before making changes.

preview_pipeline previews the tabular data in a pipeline. By default it shows the final state after all transformations. When you supply a transformation ID, it shows the data up to and including that step, which is useful for checking your work as you build.

transformation_insert inserts a new column, populated with either a static value or a computed formula. Set the value type to scalar to write a fixed value, or to formula to compute the value from a simple expression. The supported formulas are intentionally simple: for logic that needs many conditional rules, a Mapping Group is usually a better fit. You control where the new column lands with a position operator of first, last, before, or after; the before and after options are placed relative to a reference column you name. The value and formula fields both accept variables.

transformation_copy copies an existing column into a new column under a different name. It uses the same first, last, before, and after positioning as insert.

transformation_insert_row_numbers adds a column of sequential row numbers, such as 1, 2, 3, with the same positioning options.

transformation_generate_unique_id adds a column of 32-character alphanumeric unique IDs, one per row (a UUID v4 value with the hyphens removed). Two parameters control the output:

As with the other column tools, you can position the new ID column using first, last, before, or after relative to a reference column.

transformation_shift moves an existing column to a new position, placed relative to another column.

transformation_rename renames a column.

transformation_keep keeps only the columns you specify and drops the rest.

transformation_remove drops the columns you specify.

transformation_merge combines several columns into one, joined by a delimiter, with the option to drop the original columns.

transformation_split splits one column into several columns using a delimiter, for example turning a full-name column into first-name and last-name columns.

transformation_filter keeps only the rows that match the conditions you provide. Each condition compares a column, using an operator, against a value. The value can be a fixed value (a scalar comparator), the value of another column (a column comparator), or a percentile such as the top 20 percent (a percentage comparator). Conditions are grouped, and the group's combinator (and or or) determines whether every condition must pass or just one. Groups can be nested to build compound logic such as "A and (B or C)." The available operators are:

transformation_lowercase converts the values in a column to lowercase.

transformation_uppercase converts the values in a column to uppercase.

transformation_replace performs a find-and-replace on a string column. It supports two replacement types:

transformation_parse_date converts date values from one format to another and can convert time zones. You provide an input format that describes the existing values and an output format for the result, both written as standard chrono strftime codes. The output format defaults to %Y-%m-%d %H:%M:%S if you do not specify one. Invalid or empty values become null. Common format codes:

transformation_flip_sign multiplies the values in a numeric column by negative one, so positives become negatives and negatives become positives. Two capabilities extend the basic behavior:

transformation_math_operation runs an aggregate calculation across a column and writes the result to a new column at a decimal precision you choose. The supported functions are sum, count, max, min, mean, median, mode, standard_deviation, and n_distinct.

transformation_group_by groups rows by one or more columns and applies one or more aggregations, collapsing the data to one row per group (similar to GROUP BY in SQL). You can define several aggregations in a single step, each with its own source column, aggregation function, output column, and decimal precision. The available functions are sum, count, max, min, mean, median, mode, standard_deviation, and n_distinct.

join_dataset joins another dataset or supporting pipeline to the pipeline's primary data on matching columns. You choose the join type (inner, left, right, or outer) and the columns to join on, and column types are auto-cast for compatibility by default.

stack_dataset stacks the rows of another dataset or supporting pipeline on top of the pipeline's current data, appending records.

mapping_transformation creates or edits a Mapping Group and applies it to an existing column. Each rule is a find-and-replace defined by a match type, the value to match, and the result value to write when it matches. Rules run in order, and once a rule matches a value that value is not changed again, so place the most specific rules first. A Mapping Group transforms values in place rather than creating a new column, so rename the column afterward if you want a different name. The supported match types are:

mapping_group_lookup finds Mapping Groups by name within an organization.

execute_workflow triggers a manual, background execution of a workflow. The run processes asynchronously, so the request returns execution details rather than waiting for completion.

workflow_execution_lookup looks up the status and details of a single execution, including state, timing, and any errors or warnings.

all_workflow_executions lists recent executions for a workflow, with optional filters for status, start date, and result limit.

fetch_workflow_run_statistics returns execution statistics for a workflow's runs, including timing, status, and the complete statistics for each execution.

fetch_pipeline_run_statistics returns the same level of detail for an individual pipeline's executions.

get_pipeline_edit_attribution returns the last editor and edit timestamp for a pipeline, plus recent version snapshots with their associated execution IDs.

get_mapping_group_edit_attribution returns edit history for a Mapping Group, showing property changes, added, changed, and deleted mapping rules, and column changes across recent versions.

delete_transformation removes a transformation from a pipeline and returns the updated pipeline state, so you can revise a transformation chain step by step. To change a step in place rather than remove it, pass its transformation ID to the relevant transformation tool instead.

list_variables lists all variables in a workspace with their ID, name, type, and value.

update_variable updates the value of a workspace variable.

delete_variable removes a workspace variable.