Prophecy Conversion Matrix

PlaidCloud converts Prophecy Spark pipelines and SQL projects into native Advanced workflows. The importer maps each Prophecy gem to a workflow step, project variable, or canvas object, so a converted workflow runs cloud-native rather than as an emulation.

Every gem, connector and project structure converts: Full means the gem becomes a native PlaidCloud step, variable, or canvas object that does the gem’s work. The few behaviours a converted workflow does not reproduce on its own are listed under Known Edge Cases, each with what to do.

To run a conversion, see Migrate Prophecy Projects.

Spark pipeline gems

Source/Target

Gem	Status	PlaidCloud Equivalent	Converts
Lookup	Full	Lookup	Each lookup call becomes a Lookup step in the step that uses it; a key with no match reads NULL, and a duplicated key returns one of its rows.
Source	Full	Import step chosen by the dataset format (see Connectors)	Each Source becomes an import step into a table named after the dataset; a format the converter does not recognise is read as a CSV file uploaded to the workflow’s document folder, with a warning.
Streaming Source	Full	Same import as batch Source	Streaming sources convert to a one-shot batch import with a warning.
Streaming Target	Full	Batch append load	Streaming targets convert to a batch append with a warning.
Target	Full	Table export / load step chosen by dataset format and writeMode	Each Target becomes a table load or export step in the gem’s write mode: append, overwrite, merge with its SET expressions, upsert, SCD2, replaceWhere and insertInto.

Prepare

Gem	Status	PlaidCloud Equivalent	Converts
Data Cleansing	Full	Column expressions in a Table Extract step	Converts null fill, trim and case options into column expressions; cleaned columns move last.
Sequence	Full	Table Extract with a row number expression	Adds a sequence column from the start value and increment.
Surrogate Key Generator	Full	Table Extract with a row number expression	Adds a sequential key from the initial value; keys continue across runs from a table that holds the last key issued.

Transform

Gem	Status	PlaidCloud Equivalent	Converts
Aggregate	Full	Table Extract with aggregation; Pivot when the gem pivots	Groups and aggregates, applies the HAVING condition, and pivots into one column per pivot value; propagate-all carries every other input column.
Bulk Column Expressions	Full	Expand to one expression per column in a Table Extract step	One expression per listed column in a Table Extract step; the conversion warns that the gem’s layout is assumed.
Bulk Column Rename	Full	Column rename in a Table Extract step	One rename per listed column in a Table Extract step; the conversion warns that the gem’s layout is assumed.
Deduplicate	Full	Table Extract with distinct or a row-number filter	Keeps one row per key: distinct, first or last by your sort order, any, or only the keys that occur once.
Dynamic Replace	Full	Dynamic replace transform, configured by hand	Converts to replacement rules on the dynamic replace transform; the conversion warns that the gem’s layout is assumed.
Dynamic Select	Full	Column select in a Table Extract step resolved by type	Resolves columns by type at conversion time into an explicit select.
Filter	Full	Table Extract with a filter expression	Converts to a Table Extract step with a filter expression; NULL conditions drop the row.
Flatten Schema	Full	JSON Parse and Table Extract	Flattens nested fields into columns named with the gem’s delimiter and explodes arrays into one row per element.
Fuzzy Match	Full	Table Extract and User Defined Transform	Pairs the records, scores each match field by the gem’s method, averages the scores, and keeps the pairs at or above your threshold.
Group By Rollup	Full	Union of Table Extract aggregation steps per rollup level	Expands the rollup into one aggregate per grouping level unioned together with NULL subtotals.
Limit	Full	Table Extract with a row limit	Converts to a row limit; config references become project variables.
Normalize	Full	Table Extract with an explode expression	Repeats each record by the length expression, computes the output columns per index, and stops at the first index where the finished condition holds.
Order By	Full	Sort on a Table Extract step	Converts to a sort with the same keys, direction and null placement.
Reformat	Full	Table Extract with column expressions	Converts to a Table Extract step with one expression per output column; unlisted columns are dropped.
Sample Rows	Full	Table Extract with a row limit or random sample	Keeps a random fraction or count of rows; the rows chosen differ from the Spark run because sampling is random.
Schema Transform	Full	Chain of Table Extract steps applied in order	Applies each operation in order: add or replace a column, drop, rename, add if missing with its default, and add a project business rule.
Set Operation	Full	Union All; Anti Join for except; Inner Join for intersect	Union all stacks the inputs; except all and intersect all keep the right number of duplicate rows.
TransposeAgg	Full	Table Extract with aggregation	Converts the gem’s grouping, aggregates and pivot as an Aggregate, with a warning to review it.
Unpivot	Full	Melt	Stacks the selected columns into name and value columns.
Watermark	Full	No step needed	Dropped: a batch run reads all the data present, so there is no late-data cutoff to apply.
Window Function	Full	Window expressions in a Table Extract step	Adds window columns with the gem’s partition, order and frame, rows or range, with unbounded, current-row or offset bounds.

Join/Split

Gem	Status	PlaidCloud Equivalent	Converts
Compare Columns	Full	Full outer join plus per-column aggregates	Compares the columns through a full outer join of the unpivoted data, with match and mismatch counts.
Join	Full	Inner Join, Multi-Table Join, Anti Join or Cross Join	Builds the join of each type, inner, left, right, full, semi, anti, cross and natural, on any join condition, applies the where clause, and writes the select list.
Repartition	Full	No step needed	Dropped: PlaidCloud manages partitioning, so the next step reads the upstream table directly.
Row Distributor	Full	One Table Extract filter per output	Converts to one filter per output; rows can land in multiple outputs.

Parse

Gem	Status	PlaidCloud Equivalent	Converts
Column Parser	Full	JSON Parse / XML Parse	Converts to a JSON or XML parse of the named column; the conversion warns that the gem’s layout is assumed.

Custom

Gem	Status	PlaidCloud Equivalent	Converts
Custom Gem (Gem Builder)	Full	User Defined Transform running the gem’s generated code	The gem’s generated PySpark code from the pipeline export runs in a Python step, with a warning to review it.
CustomReformatBitbucket	Full	Table Extract	Converts as the Reformat whose expressions the gem carries, with a warning to review it.
Data Quality Check	Full	Union All of one result row per check; Raise Workflow Error to stop on failures	Evaluates each check over the input and writes one result row per check with its status and message. The data passes through, and the run stops when the gem’s failure threshold is reached.
Deequ	Full	Table Extract checks	Passes data through and builds the constraint results table for the isComplete, isUnique, isNonNegative and isContainedIn rules.
Delta Table Operations	Full	Table In-Place Delete, Table Update, Table Drop, Table Restore to Timestamp	Delete, update, drop and restore to a timestamp map to table steps; optimize, vacuum and restore by version pass data through with a warning.
Directory	Full	Directory Listing	Lists the files of the bound document folder in the gem’s columns; with the sheet-name option, each workbook gives one row per sheet. A missing folder gives zero rows.
Email	Full	Notify Via Email	Sends the email, with an optional data attachment, through a bound email connection.
Email Data	Full	Notify Via Email	Emails the input rows as a table and passes the input through unchanged.
File Operation	Full	Copy Document File or Copy Document Directory	Copies, moves or syncs files from the bound document folder into the converted workflow’s document folder; a move removes the source files.
Log	Full	Table Extract	Writes the gem’s event row (time, gem name, event type and text) to its output table.
REST API Enrich	Full	REST Request	Calls the API once per row and adds the response columns.
Script	Full	Python step with frame-access rewrite	Embeds the script body in a python step, with data-frame access rewritten; Scala bodies carried with a note.
Soda Data Quality Check	Full	Table Extract checks	Builds the check results table from the Soda checks.
SQL Statement	Full	SQL step per output tab, text translated to the tenant dialect	Embeds each tab as a SQL step with inputs bound as temp tables; translated to the tenant dialect.
Todo	Full	Raise Workflow Error or Notify via Log	Marks logic still to be written: a log message when nothing reads it, or a step that stops the run with the gem’s message when a later step depends on it.

Report

Gem	Status	PlaidCloud Equivalent	Converts
Tableau	Full	Publish to Tableau	Writes a Hyper extract and publishes it to the bound Tableau connection.

ML/AI

Gem	Status	PlaidCloud Equivalent	Converts
Bedrock	Full	Text NLP AI	Embeds the text column through the workspace AI connection.
OpenAI	Full	Text NLP AI	Answers questions or embeds text through the workspace AI connection.
Pinecone Lookup	Full	REST Request	Queries the Pinecone index for each row and adds pinecone_matches and pinecone_error columns.
SageMaker Endpoint	Full	Text NLP AI	Answers questions through the workspace AI connection.
Text Processing	Full	Table Extract for chunk splitting; REST Request for web scraping	Chunk splitting becomes a character-chunk expression; web scraping fetches each row’s URL and joins the page content back onto the row.

Subgraph/Control

Gem	Status	PlaidCloud Equivalent	Converts
Control Flow Input	Full	Dropped (boundary marker only)	Dropped; it only marks the iterator sub-workflow input boundary.
Control Flow Output	Full	Dropped (boundary marker only)	Dropped; it only marks the iterator sub-workflow output boundary.
Subgraph (Basic)	Full	Canvas container / nested workflow from inline processes	Converts the inline inner graph into a nested workflow (or inlined steps) with in/out table parameters.
Table Iterator	Full	Workflow Loop	Runs the loop body once per input row, setting each configuration value from its column, up to maxIteration.
While Iterator	Full	Workflow Loop	Runs the loop body once per pass, passing the iteration number, until the body returns no rows or the gem’s maximum is reached.

SQL / Analyst gems

Source/Target

Gem	Status	PlaidCloud Equivalent	Converts
External table Source/Target (BigQuery, Databricks, SAP HANA, MongoDB, MSSQL, Oracle, Postgres, Redshift, Salesforce, Snowflake, Synapse)	Full	External Data import/export step with the matching database connection	Reads or writes a table in an external database through a PlaidCloud connection.
File storage Source/Target (ADLS, Databricks Volumes, GCS, OneDrive, S3, SFTP, SharePoint, Smartsheet)	Full	Import/Export file steps against a document account (ADLS/GCS/S3/SFTP/SharePoint/OneDrive)	Reads or writes files in cloud/SFTP storage with wildcard union, trigger, delete/move and encryption options.
Model	Full	Upstream step output table	Reads another model as the table produced by the referenced upstream step.
Seed	Full	Import: CSV	Seed CSV is imported into a PlaidCloud table.
Source	Full	Import step chosen by the dataset format (see Connectors)	Each Source becomes an import step into a table named after the dataset; a format the converter does not recognise is read as a CSV file uploaded to the workflow’s document folder, with a warning.
Target Model / Table (Target)	Full	Table Extract, Upsert, Copy Table and Union All	Writes the model’s final select to a PlaidCloud table as replace, append, upsert, delete-and-insert, SCD2 or replace by predicate.
Upload Files	Full	Import: CSV	Uploaded local files become a table via an import step.

Prepare

Gem	Status	PlaidCloud Equivalent	Converts
Data Cleansing	Full	Column expressions in a Table Extract step	Converts null fill, trim and case options into column expressions; cleaned columns move last.
Data Masking	Full	Column expressions for character masking or hashing	Masks or hashes selected string columns in place or as new columns.
Find Duplicates	Full	Row-number window expression plus filter	Returns unique, duplicate or count/position-selected rows per key.
Generate Rows	Full	Sequence table joined and filtered per input row	Generates rows per input row from init, condition and loop expressions.
Imputation	Full	Aggregate plus join-back and column expression	Replaces nulls or sentinel values with average, median, mode or a fixed value.
Multi-Column Edit	Full	Per-column expressions expanded from the template	Applies one expression template to every selected column.
Multi-Column Rename	Full	Column rename mapping on a select step	Renames selected columns by prefix, suffix or name expression.
Record ID	Full	Row-number or uuid column expression	Adds a UUID or sequential record id column.
Sample	Full	Row limit or sample option on extract	Takes first/last/random/Nth/percent rows overall or per group.
Tile	Full	Window expressions for tile number and sequence	Assigns tile numbers by equal sum, count, std-dev bands, distinct values or manual cutoffs.

Transform

Gem	Status	PlaidCloud Equivalent	Converts
Aggregate	Full	Table Extract with aggregation; Pivot when the gem pivots	Groups and aggregates, applies the HAVING condition, and pivots into one column per pivot value; propagate-all carries every other input column.
Count Records	Full	Aggregate step with count functions	Returns total, non-null or distinct counts as one row.
Data Encoder/Decoder	Full	Column expressions for base64/hex encode and decode	Encodes or decodes selected columns; AES only on Databricks.
Deduplicate	Full	Table Extract with distinct or a row-number filter	Keeps one row per key: distinct, first or last by your sort order, any, or only the keys that occur once.
Dynamic Select	Full	Column select in a Table Extract step resolved by type	Resolves columns by type at conversion time into an explicit select.
Filter	Full	Table Extract with a filter expression	Converts to a Table Extract step with a filter expression; NULL conditions drop the row.
Finance	Full	Column expression implementing the financial function	Appends a financial function result (FV, PV, NPV, IRR, …) as a new column.
Flatten Schema	Full	JSON Parse and Table Extract	Flattens nested fields into columns named with the gem’s delimiter and explodes arrays into one row per element.
Fuzzy Match	Full	Table Extract and User Defined Transform	Pairs the records, scores each match field by the gem’s method, averages the scores, and keeps the pairs at or above your threshold.
Limit	Full	Table Extract with a row limit	Converts to a row limit; config references become project variables.
Order By	Full	Sort on a Table Extract step	Converts to a sort with the same keys, direction and null placement.
Pivot	Full	Pivot	Turns the selected pivot values into columns of aggregates; a Simple-mode value list fetched at design time is converted as the listed values.
Reformat	Full	Table Extract with column expressions	Converts to a Table Extract step with one expression per output column; unlisted columns are dropped.
Running Total	Full	Cumulative-sum window expression column	Appends cumulative sums per selected column.
Snowflake Unpivot (third-party)	Full	Melt	Unpivots columns via Snowflake UNPIVOT.
Transpose	Full	Melt	Stacks data columns into name and string value columns.
TransposeAgg	Full	Table Extract with aggregation	Converts the gem’s grouping, aggregates and pivot as an Aggregate, with a warning to review it.
Unpivot	Full	Melt	Stacks data columns into name and value columns.
Weighted Average	Full	Aggregate step with weighted-average expression	Computes a weighted mean overall or per group.
Window Function	Full	Window expressions in a Table Extract step	Adds window columns with the gem’s partition, order and frame, rows or range, with unbounded, current-row or offset bounds.

Join/Split

Gem	Status	PlaidCloud Equivalent	Converts
Except	Full	Anti Join on all columns plus a distinct Table Extract	Keeps rows of the first input that appear in none of the other inputs.
Intersect	Full	Inner Join on all columns plus Distinct	Keeps rows present in every input.
Join	Full	Inner Join, Multi-Table Join, Anti Join or Cross Join	Builds the join of each type, inner, left, right, full, semi, anti, cross and natural, on any join condition, applies the where clause, and writes the select list.
Union	Full	Union All	Stacks the inputs by position, removing duplicates when the gem does not keep them.
Union By Name	Full	Union All	Becomes a Union All that aligns columns by name and null-fills missing ones.

Parse

Gem	Status	PlaidCloud Equivalent	Converts
JSON Parse	Full	JSON Parse	Parses a JSON text column into structured fields.
Regex	Full	Regex expressions in a Table Extract / Transform step	Replace, parse, match or tokenize a text column with a regex.
Text To Columns	Full	Split-column / explode steps with regex delimiter	Splits a delimited column into columns or rows using a regex delimiter.
XML Parse	Full	XML Parse	Parses an XML text column into structured fields.

Custom

Gem	Status	PlaidCloud Equivalent	Converts
Directory	Full	Directory Listing	Lists the files of the bound document folder in the gem’s columns; with the sheet-name option, each workbook gives one row per sheet. A missing folder gives zero rows.
DynamicInput	Full	SQL step for query templates; workflow loop for multi-file mode	Query-template mode embeds the template as a SQL step; multi-file mode maps with assumptions.
Email	Full	Notify Via Email	Sends the email, with an optional data attachment, through a bound email connection.
RestAPI	Full	REST Request	Sends one request per input row with your URL, headers, query, body and retries, keeps the response, and polls an asynchronous request until it is done. Credentials come from the connection you bind.
Script	Full	Python step with frame-access rewrite	Embeds the script body in a python step, with data-frame access rewritten; Scala bodies carried with a note.
SQL Statement	Full	SQL step per output tab, text translated to the tenant dialect	Embeds each tab as a SQL step with inputs bound as temp tables; translated to the tenant dialect.
StoredProcedure	Full	Agent Remote Execution of SQL	Calls the procedure on its warehouse through the connection you choose: once per input row with each argument bound from its column, or once when it takes no arguments. The input columns pass through.
Table Operations	Full	Table In-Place Delete, Table Update, Table Drop	Update, delete and drop map to table steps; vacuum and optimize become table optimize or a no-op with a warning.
ToDo	Full	Raise Workflow Error or Notify via Log	Marks logic still to be written: a log message when nothing reads it, or a step that stops the run with the gem’s message when a later step depends on it.

Spatial

Gem	Status	PlaidCloud Equivalent	Converts
Buffer	Full	Spatial Buffer	Grows each geometry by the configured distance in miles or kilometers.
CreatePoint	Full	Table Extract with geom_point	Builds point geometry from longitude and latitude columns.
Distance	Full	Table Extract	Computes distance, cardinal direction and bearing between two WKT points.
FindNearest	Full	Table Extract with a cross join	Finds N nearest target points for each source point within a max distance.
HeatMap	Full	Heat Map (macro)	Bins points into H3 hexagons at the gem’s resolution, spreads the heat over the grid distance with its decay, and gives each hexagon’s density and outline.
PolyBuild	Full	Table Extract and a Python step	Builds polygon or polyline WKT from ordered points per group.
Simplify	Full	Spatial Generalize	Simplifies WKT geometries by a tolerance in miles/kms.
SpatialMatch	Full	Spatial Match	Matches source and target geometries by intersects/contains/within/touches/envelope.

Report

Gem	Status	PlaidCloud Equivalent	Converts
PowerBIWrite	Full	REST Request	Creates the Power BI dataset and pushes the rows through the Power BI REST API.
TableauWrite	Full	Publish to Tableau	Writes a Hyper extract and publishes it to the bound Tableau connection.
Visualize	Full	Copy Table	Passes the rows on unchanged; build the chart as a PlaidCloud dashboard on the same table.

Subgraph/Control

Gem	Status	PlaidCloud Equivalent	Converts
Condition	Full	Cross Join and Table Extract	Evaluates the conditions over the whole input and sends every row to the first output whose condition holds, else to the last output.
Macro	Full	SQL step with the macro body expanded	Embeds the macro body as a SQL step translated to the tenant dialect, inputs bound as temp tables.
Pipeline (trigger)	Full	Workflow Loop	Runs the named pipeline once per input row with your parameters and trigger condition, continues past a failed run, and returns one row per run.
Subgraph (Basic)	Full	Canvas container / nested workflow from inline processes	Converts the inline inner graph into a nested workflow (or inlined steps) with in/out table parameters.

Connectors

Dataset formats and connectors are source and target settings of a gem, not gems. Each format converts to the import or export step shown, and conversion never contacts the source or target system.

Format	Status	PlaidCloud Equivalent	Converts
Avro	Full	Import: Avro; Export: Avro	Reads and writes Avro files.
BigQuery	Full	External database connection with a table import	Reads and writes through a connection.
Binary files	Full	Directory Listing with file contents; Alteryx Executor to write	Reads each file of the folder with its path, modification time, length and content; writes each row’s content to the file its path names.
Catalog table	Full	Existing PlaidCloud table binding	Binds to a table by name; reads and every write mode convert.
Cosmos DB	Full	Import and export: REST Request	Reads and writes the container’s documents through your Cosmos DB connection.
CSV	Full	Import: CSV; Export: CSV	Reads and writes delimited files with header, separator, quote, escape, and encoding options.
Data Generator	Full	Table Faker	Generates the dataset’s columns with the same kinds of fake data.
DataHub table	Full	Import SQL	Reads the table the dataset names through your connection; a write lands in a PlaidCloud table.
Delta	Full	Existing PlaidCloud table binding	Reads and writes the table in every write mode; a read of an earlier version reads it through the Databricks connection you bind.
Excel (XLSX)	Full	Import: Excel; Export: Excel	Reads and writes workbook sheets and ranges.
Fixed format	Full	Import: Fixed Width; Export: CSV	Reads fixed-width files laid out from the dataset’s column widths; writes a delimited file.
Hive	Full	Import: Import SQL; Export: External Tables	Reads and writes the table through your Hive connection.
Iceberg	Full	Existing PlaidCloud table binding	Reads and writes the table.
JDBC	Full	External database connection with a table import	Reads a table or query and writes in replace or append mode through your connection; a post-write statement runs after the write.
JSON	Full	JSON Parse	Loads records and flattens nested fields.
Kafka	Full	Import: Import Kafka; Export: Export Kafka	Reads a topic as one batch of messages and writes rows as messages.
MongoDB	Full	Import: MongoDB; Export: MongoDB	Reads and writes the collection through your MongoDB connection.
ORC	Full	Import: ORC; Export: ORC	Reads and writes ORC files.
Parquet	Full	Import: Parquet; Export: Parquet	Reads and writes Parquet files.
Pinecone	Full	Import and export: REST Request	Reads an index’s records and upserts rows as vectors through the Pinecone REST API.
Redshift	Full	External database connection with a table import	Reads and writes through a connection.
Salesforce	Full	Import: Query Salesforce Object; Export: REST Request	Reads an object or SOQL query and writes records through your Salesforce connection.
Seed (inline)	Full	Import: CSV	Loads the inline rows as a table.
Slack	Full	Import: Singer; Export: REST Request	Reads channel messages and posts each row as a message through your Slack connection.
Smartsheet	Full	Import and export: REST Request	Reads and writes the sheet’s rows through your Smartsheet connection.
Snowflake	Full	External database connection with a table import	Reads and writes through a connection.
SQL file types (CSV, Excel, JSON, Parquet, Text, XML, fixed-width)	Full	Import and export file steps against a document account	Reads and writes the file type; options without a PlaidCloud equivalent are flagged.
Text	Full	Import: CSV as a single column	Reads each line as one column.
Web page	Full	REST Request	Reads the page at the dataset’s URL into a table.
XML	Full	XML Parse	Loads records by row tag.

Expressions and Project Structure

Area	Status	PlaidCloud Equivalent	Converts
Expressions	Full	Spark and warehouse SQL expressions translated function by function	Functions and operators convert to PlaidCloud expressions; a function without an equivalent is flagged by name.
Project and graph structure	Full	Project loader	Every Prophecy export shape loads and normalizes into one model.

Known Edge Cases

Case	What happens
Code or data the export does not include	A Gem Builder or other custom gem whose code is not in the export, and a call to a macro of a dbt package the export leaves out, become a step that names what is missing; export the project with its pipeline code and dbt packages, or rebuild the logic in a Python or SQL step. Ab Initio Error Detection and MTime gems pass their input through, because the rule they apply is not in the export. Soda checks read from a file outside the project, and Soda checks that compare with earlier scans or with another dataset or test a format Soda names, have no row in the results table; compute them in a later step.
Results only known when the pipeline runs	A REST API gem that flattens its response puts the whole response in one column when nothing declares the field names, which only a run reveals; extract the fields you need in a later step. A dbt macro that queries the warehouse as dbt compiles becomes a step that names it; replace the query with the values it returns. A Spark StoredProcedure gem runs, but its OUT values are not returned; the output is the pass-through columns. input_file_name() returns an empty string over a table or Delta source, which has no file to name, and hash() / xxhash64() inside an aggregate, a window or a gem with several inputs are computed by the warehouse, so they can differ from Spark’s.
Values written into a step when the project converts	A parameter value holding a quote, backslash or brace, a pipeline value a SQL step reads, and the database a Delta table operation names through configuration are written into the step as fixed values, so changing the workflow variable does not change them; edit the step when the configuration changes.
Scala code	A Scala script, Scala expression, or Scala user-defined function or aggregate function becomes a step that names it; rewrite it in SQL or Python.
Kafka Kerberos sign-in	A Kafka source or target that signs in with Kerberos (GSSAPI) converts without the Kerberos service name and sign-in mechanism; set them on the PlaidCloud connection.
Streaming	Streaming sources, targets and pipelines run as batch workflows over the data present when they run; schedule the workflow to run as often as you need.
Job components	A Spark job converts its one pipeline and schedule; its other components, run conditions and configuration overrides are not converted.
Uncommon calls	A user-defined aggregate function called from a rollup or a window, and a dbt macro that uses Jinja beyond if, for, set and the common filters, become a step that names the gem; call the function from an Aggregate gem, or inline the macro.

Validating a Converted Workflow

Compare the converted workflow’s output against the output of your own Prophecy pipeline: schema, row count, and row values. Ignore row order unless the workflow depends on it.