Managing Tables and Views
PlaidCloud offers the ability to organize and manage tables, including labels. Tables are available to all workflows within a project and have many tools and options.
In addition to tables, PlaidCloud also offers Views based on table data. Using Views allows for instant updates when underlying table changes occur, as well as saving data storage space.
Options include:
- The same table can exist on multiple paths in the hierarchy (alternate hierarchies)
- Tables are taggable for easier search and inclusion in PlaidCloud processes
- Tables can be versioned
- Tables can be published so they are available for Dashboard Visualizations
PlaidCloud uses a path-based system to organize tables, like you would use to navigate a series of folders, allowing for a more flexible and logical organization of tables. Using this system, tables can be moved within a hierarchy, or multiple references to one table from different locations in the hierarchy (alternate hierarchies), can be created. The ability to manage tables using this method allows the structure to reflect operational needs, reporting, and control.
Finding Tables
Section titled “Finding Tables”Use Filter at the bottom of the list, beside the row count, to search table names and labels. Press / or Ctrl/Cmd+F to jump to it. The same filter applies to the tree and to the flattened list. The All | Published switch beside the filter narrows the list to published tables; it is set to All by default and applies in both the tree and the flat list.
By Workflow
Section titled “By Workflow”The By Workflow folder at the end of the Tables tab groups tables under each workflow that reads or writes them. Open a workflow to see its tables. The folder is a view of the same tables, so you can’t move or copy anything from it.
By Column
Section titled “By Column”Type a column name in the column search box on the Tables tab and press Enter, or select the magnifier beside it, to show only the tables that have a column with that name. The search reads column names, not the data, so it is fast. Select Contains to match part of a column name instead of the whole name. Clear the box to show every table again. A long search can return in pages and the status line says when it is still searching or may be incomplete.
Select Column Tree to search from a window of its own. Each table that has the column opens to two lists:
- Fed by — the tables that supply the column
- Feeds — the tables the column goes on to feed
An AI assistant connected over MCP finds the same tables with table_find(column=...), matching the whole name or part of it.
Search Values
Section titled “Search Values”To find which tables hold a value, choose Search values in data… (slow) from the Actions menu on the Tables tab, or right-click a table and choose it to look in that table only. Enter the value, choose Contains to match part of a value, and select Search. Each table that holds the value is listed with the columns it appears in and a few of the matching values. Double-click a table, or press Enter on it, to open it in the Table Explorer. A column opens the table filtered to the rows that match. If the column search box holds text, it fills in the value to find.
Before you search, know that:
- The search reads every row of every table searched, so it can be very slow and use a lot of warehouse resources. The window warns you, and the search buttons stay off until a search finishes.
- You see only rows you could already open. Row-level security applies to every search.
- Your warehouse’s query history records the values you search for.
- The search covers the text columns of tables. Tables that haven’t been built are skipped and named in a warning.
- Results come in pages. When the status line says the search stopped early, choose Search more to continue with the tables not yet searched.
An AI assistant connected over MCP uses table_value_search. It returns the tables and columns that hold the value, never the values themselves. When you ask the AI assistant in a project to find a value, Weave sends the names of the matching tables and columns to the language model provider.
Import a CSV File
Section titled “Import a CSV File”To turn a CSV file into a new table in this project:
- Drag the file from your computer onto the project’s Tables list, or choose New > Table from CSV File on the toolbar and drop the file onto the box or click Choose a CSV file.
- PlaidCloud uploads the file to your Personal Temp Drive, or to the workspace’s Temp Drive if you don’t have one, and imports it into a new table at the top level of the list, named after the file:
Sales Data.csvbecomessales_data. If the project already has a table with that name, the new one gets a number, such assales_data_2. - The list refreshes with the new table selected, and Your data at a glance opens on it. See Start With Your Own CSV for what the panel shows.
Only files ending in .csv are accepted. You need permission to create tables and to write to document storage, and a locked project doesn’t accept a CSV until it is unlocked. If the import fails, the window says why and your project is left as it was. PlaidCloud deletes the uploaded file from the temp drive once the import has finished, whether or not it succeeded; failing that, the temp drive deletes it after a day.
To move a table:
- Drag it into the folder where you wish it to be located
To move a table to the top level of the project, drag it onto the empty space below the last row. In folder pickers, / is the top level of the project.
Rename
Section titled “Rename”To rename a table:
- Right click on the table
- Select the rename option
- Type in the new name and save it
- The table is now renamed, but it retains its original unique identifier.
To clear a table:
- Select the tables in the hierarchy ‘
- Click the clear button on the top toolbar.
Note: You can clear a single table or multiple tables
Delete
Section titled “Delete”To delete a table:
- Select the tables in the hierarchy
- Click the delete button on the top toolbar
- The deleted operation will check to see if the table is in use by workflow steps or Views. If so, you will be asked to remove those associations before deletion can occur.
Note: You can also force delete the table(s). Force deletion of the table(s) will leave references broken, so this should be used sparingly.
Last Read
Section titled “Last Read”The delete window shows when the table was last read, for example Last read 3 days ago by a person or step, so you can tell a table nobody uses from one that is still in use. PlaidCloud takes this from your warehouse’s own query history and leaves out reads made by its own housekeeping.
| Warehouse | What is needed |
|---|---|
| Snowflake | Reads come from Access History, which Snowflake records up to about 3 hours late. The connection’s role needs the SNOWFLAKE.GOVERNANCE_VIEWER database role (IMPORTED PRIVILEGES on the SNOWFLAKE database also works). Snowflake Standard edition has no Access History. |
| Databricks | Reads come from the lineage table, which Databricks also records with a delay, so a recent read may not show yet. The connection’s identity needs USE CATALOG on system, USE SCHEMA on system.access, and SELECT on system.access.table_lineage. |
| PlaidCloud Lakehouse | Read history is not yet available. The window shows Read history incomplete. |
When you connect your warehouse, the connection check recommends any of these grants that are missing, each with the statement to run. They are recommended, not required: without them the warehouse connects and the window shows Read history incomplete.
When PlaidCloud can’t read the whole history, the line says Read history incomplete and never reports “no reads” in its place. Anyone with direct SQL access to your warehouse can set a session tag that changes how their own reads are labelled.
An AI assistant connected over MCP gets the same answer from table_describe(include=['last_read']).
See a Table’s Columns
Section titled “See a Table’s Columns”Selecting a table opens its details panel. Its Columns section lists each column with its Type and Description, the description coming from Column Properties. Point at a row to read a long description in full.
A table that isn’t in the warehouse right now, because its data expired or no step has built it, has no columns to list. The section says so and shows how many columns the table has. Run a step that writes the table to list them.
See Where a Table Is Used
Section titled “See Where a Table Is Used”Selecting a table opens its details panel, which lists the workflow steps that touch it:
- Table Created By — steps that write to the table
- Table Contributes To — steps that read from it
- May Reference This Table — steps that reach a table through a variable that cannot be resolved. This section only appears when there are such steps.
Selecting another table clears these lists straight away. A step in another project that shares this project’s data, such as an Import: External Project Table step reading this table, is listed too, without its name or workflow, because you may not have access to that project. Those steps can appear a moment after the rest.
A step can name a table outright, or reach it through a path containing a variable, such as /Periods/{period}/Sales. Where that variable has a current value, the step appears in the first two lists with an information icon beside it. Hover the row to see the path as the step stores it.
Treat those rows as the weaker answer they are. They describe what a run starting now would touch — not everything that has ever touched the table. Change the variable and a different table is read.
Some variables cannot be resolved. Either nothing has set the variable at all, or the value a run uses does not come from what is stored — a Workflow Loop takes its variables from data as it iterates, a Set Variable step assigns during the run, a macro is handed its variables by whichever step calls it, and a run started by a sensor carries that sensor’s own details. In each case whatever is stored between runs is left over from the last one rather than a value to trust. Rather than resolve those to a table that may be the wrong one, PlaidCloud lists the steps by name under May Reference This Table, with the variable it could not resolve, so you can open them and check.
A step your role can’t see is still listed, so you know the table is in use, but without its name or workflow, and you can’t open it. It still counts when you delete the table. A step you can see is listed without any workflow your role hides. Nothing is hidden from an Architect.
Note: When you delete tables from the Tables list, a step that reaches a table through a path, with or without a variable, counts as using it, even when its variable can’t be resolved. The delete window lists the step, and the delete goes ahead only if you force it.
Create New Directory Structure
Section titled “Create New Directory Structure”To add a new folder:
- Click the New Folder button on the toolbar
To add a folder to an existing folder:
- Right-click on the folder
- Select New Folder
When the selection is inside a folder, the prompt for the new folder’s name also has Create at the top level. Tick it to put the new folder at the top level of the project instead.
View Data (table Explorer)
Section titled “View Data (table Explorer)”Table data is viewed using the Data Explorer. The Data Explorer provides a grid view of the data as well as a column by column summary of values and statistics. Point-and-click filtering and exporting to familiar file formats are both available. The filter selections can also be saved as an Extract step usable in a workflow.
Publish Table for Reporting
Section titled “Publish Table for Reporting”Dashboard Visualizations are purposely limited to tables that have been published. When publishing a table, you can provide a unique name that may distinguish the data. This may be useful when the table has a more obscure name on part of the workflow that generated it, but it needs a clearer name for those building dashboards.
Published tables do not have paths associated with them. They will appear as a list of tables for use in the dashboards area.
Mark Table for Viewing Roles
Section titled “Mark Table for Viewing Roles”The viewing of tables by various roles can be controlled by clicking the Explorer or Manager checkboxes. If multiple tables need to be updated, select the tables in the hierarchy and select the desired viewing role from the Actions menu on the top toolbar.
Memos to Describe Table Contents
Section titled “Memos to Describe Table Contents”Add a memo to a table to help understand the data.
View Table Shape, Size, and Last Updated Time
Section titled “View Table Shape, Size, and Last Updated Time”The number of rows, columns, and the data size for each table is shown in the table hierarchy. For very large tables (multi-million rows) the row count may be estimated and an indicator for approximate row count will be shown.
View Additional Table Attributes
Section titled “View Additional Table Attributes”To view and edit other table attributes:
- Select a table
- Click the view the table context form on the right.
Duplicate a Table
Section titled “Duplicate a Table”To duplicate a table:
- Selecting the table
- Click on the duplicate button on the top toolbar.
Create a Panel App From a Table
Section titled “Create a Panel App From a Table”Select a table or view and choose Create Panel App from the New menu or the right-click menu. The New Studio App dialog opens with the project and the table already chosen. The app takes the table’s name; change it if you like, then choose Create, and Panel Studio opens it with Weave ready to build its first page. The command is available when you select a single table or view, have edit access to the project, and the project isn’t read-only. See Creating a Panel App.
See What Tables Store
Section titled “See What Tables Store”The Tables list shows Current data, History, Time Travel Days and Note columns beside each table’s size. Time Travel Days is the table’s own setting; 1 means it uses the project’s default. The status bar below the list gives the project’s total: how much it stores, and how much of that is current data and how much is history. Sizes marked Estimated come from the table’s shape until PlaidCloud measures it.
Choose Storage from the Actions menu to open the project’s Storage window, with suggestions for freeing space and the time travel settings. See Managing Storage.
Create a Data Editor from a Table
Section titled “Create a Data Editor from a Table”You can turn a table into a configured data editor in one step — select the table and choose Create Data Editor from Table from the New menu or the right-click menu. PlaidCloud examines the table’s columns and pre-builds the editor’s input controls. See Create a Data Editor from a Table for details.