--flag or --no-flag. To give a list of values, repeat the flag, for example --tables summary --tables table. The JSON examples are written for bash; on Windows you can also pass a JSON file with --input.
The commands for signing in and managing the connection are on their own pages: Sign in (login, logout, status), API keys (token) and Inspect commands on this page (tools list, tools describe). expl server skill prints a description for AI agents in the form of a SKILL.md file.
Command syntax index
Find and read
Data
Charts and analytics
Connection and management
Inspect commands
Inspect the commands the server offers and their flags.List the commands
Describe one command
expl server tools describe "chart get-data". The tool name, such as chart_get_data, also works.
It shows what the command does and the flags it takes. Check here instead of guessing command names.
The shape of a command
search and fetch take no verb: expl server search <query> and expl server fetch <id>. Other commands put a noun and a verb after server, as in expl server dataset query. Each command also has the name of the tool behind it, the one an MCP client sees. expl server call <tool_name> runs a command by that name, for example expl server call dataframe_preview <id>. To see the usage and flags of every command, see Command syntax index.expl server search
Search the content you can view by title, description and tags. Without search words, list items newest first.Usage
Arguments and flags
Examples
Search by words
TerminalList your content
TerminalOutput and limits
Results include the id and URL of each item. Pass an id to fetch or another command. Data items also list their columns (name and type) and row count. The default output is for people;--json prints JSON.
When has_more in the result is true, pass next_offset to --offset to get the next page.
Troubleshooting
Related: expl server fetch, expl server dataframe get-schema
expl server fetch
Read one note, dashboard, slide or data item by its id.Usage
Arguments and flags
Examples
Read an item
TerminalRead one heading
Terminal--query below.
Result
Find a topic
Terminal--query "region" on the same note. Each paragraph that matches comes with its heading id and path in square brackets.
Result
Output and limits
Notes, dashboards and slides come back as Markdown with their text, the first rows of each embedded chart and the explanation of each analytics. metadata.charts and metadata.blocks hold the names of charts and analytics. Data items include columns, the row count, sample rows and the update status. For a long note, find a topic with--query or read one heading with --section. When more text follows, pass the value of metadata.range.next_cursor to --cursor. metadata.truncated is set only when more text follows.
Troubleshooting
Related: expl server search, expl server chart get-data, expl server analytics get-result
expl server dataframe get-schema
Look up the columns of a data item: names, types, descriptions the publisher wrote, and the row count.Usage
Arguments and flags
Examples
Check only column names and types
TerminalCheck in detail
TerminalOutput and limits
The result includes the column names, types and the row count. With full it also includes the descriptions the publisher wrote, missing and unique counts, minimum and maximum, top values and the update status (when the data was last refreshed and whether it is fresh). When the owner has approved them, it includes semantics: what a row means, column roles, default aggregations and aggregation rules. Semantics is the official definition, so read it before you aggregate. There is no way to page through the result.Troubleshooting
Related: expl server dataframe preview, expl server dataset query
expl server dataframe preview
Get the rows of a data item, page by page with offset and limit.Usage
Arguments and flags
Examples
See the first 3 rows
TerminalGet rows 101–200
TerminalOutput and limits
When the publisher included all data, every row can be read. Otherwise the rows come from the sample saved at publish time, so check is_sample and row_count. To compute totals or filter rows, use dataset query instead of reading every page. Pass the number of rows you have already read to offset to read on. One call returns at most 100 rows.Troubleshooting
Related: expl server dataframe get-schema, expl server dataset query
expl server dataset query
Filter and aggregate a data item on the server and return only the result. Check exact column names with search or dataframe get-schema.Usage
Arguments and flags
JSON input structure
These are JSON input templates, not results. Replace < > placeholders with data ids, columns and values. Command examples using single quotes use bash syntax. PowerShell and cmd write quotes differently from bash, so an example that uses single quotes may not run as it is. In that case, save the JSON to a file and pass it with--input.
Filter conditions: filters
JSONDatasets to join: datasets
JSONJoin conditions: joins
JSONExamples
Count rows for each region
Terminal--limit 5) cut the result, so result.has_more is true and a RESULT_TRUNCATED warning is attached. To see the other regions, raise --limit or narrow the query with a filter.
Count all rows
TerminalOutput and limits
The result is exact and includes the data version (dataset.version) and update time (dataset.updated_at). Say which version of the data the numbers are based on. A join also returns the source data, where each output column comes from, and the row counts before and after each join. There is no paging. When has_more in the result is true, narrow the filters or aggregate. To get many rows, save them as a file with dataframe export.Troubleshooting
Related: expl server dataframe get-schema, expl server dataframe export
expl server dataframe export
Save the result of a query on a data item as a CSV or Parquet file and return a download link.Usage
Arguments and flags
JSON inputs
See dataset query for the JSON structure of filters, metrics and sort.Examples
Save rows as a file
TerminalOutput and limits
A download link that expires in 15 minutes is returned. The link works only for the same signed-in user. Use--output to choose where to save a file result.
There is no paging. Use this command when the result is too large for dataset query.
Troubleshooting
Related: expl server dataset query, expl server dataframe preview
expl server dataframe get-wrangling
Look up how a published data item was built: the data wrangling steps in order, the data it was read from and joined with, and which steps created or changed each column.Usage
Arguments and flags
Examples
See the outline
TerminalSee the steps that made one column
TerminalOutput and limits
Without--steps or --column, the result is an outline: an overview, every step in order as one line (type, label and what it does), the data it was read from and joined with, and an index of which steps created or changed each column. To see a step in detail (the full explanation, the expression, the settings such as filter conditions or join keys), run it again with --steps (the refs from the outline) or with --column. The refs are valid only for the current version of the data.
The overview is a summary written by AI and not reviewed, so say so when you quote it. For the processing summary the owner has approved, read the semantics in dataframe get-schema.
An outline that does not fit in one result continues with --cursor. When the steps cannot be read, the result says why: they were not recorded when the data was published, the owner does not share them, or the publisher does not share the rows of the data.
Troubleshooting
Related: expl server dataframe get-schema, expl server dataset query
expl server chart get-data
Get the data of a chart together with what the numbers mean.Usage
Arguments and flags
Examples
Get the first 5 rows of a chart
TerminalOutput and limits
Besides the rows, the result explains the numbers: chart.encoding (column, aggregation and number format per shelf), chart.filters, chart.data_granularity (aggregated or row level), chart.unsupported (what cannot be returned) and context (publication, data time and parameter values). The numbers are the values as published; results for other parameter values in the browser are not available. A chart published on its own returns the first 100 rows of the data used to draw it, not aggregated values or the values of the whole chart, so do not add the rows up as a total. Pass the number of rows you have already read to offset to read on. One call returns at most 100 rows.Troubleshooting
Related: expl server chart query, expl server chart get-image
expl server chart get-image
Get a chart as a PNG image, to see its shape, trend and layout.Usage
Arguments and flags
Examples
Get a chart image
TerminalOutput and limits
The image is the one saved when the item was published; it is not drawn again, so it can be older than the data. Check result.image.may_be_stale. The image is saved on the computer that runs expl, and image_path in the result gives the location. Use--output to choose where to save it. Read exact numbers with chart get-data, not from the image.
There is no paging.
Troubleshooting
Related: expl server chart get-data
expl server analytics get-result
Get the results of analytics (regression, correlation, chi-square test and so on) as structured data.Usage
Arguments and flags
Examples
Get an analytics result
Terminal--analytics is not passed. Long output is trimmed. The explanation texts are left empty in this example.
Result
Output and limits
The result holds the model definition (target, predictors, N), result tables with stable column ids, headline statistics split into training and test, the settings it ran with, and explanation texts marked as generated or AI-generated. Values are as displayed, and coefficients are unstandardized. Use--limit to cap the rows of each table. There is no way to page through the rest.
Troubleshooting
Related: expl server fetch
expl server chart query
Filter and aggregate the data of a chart in a note, dashboard or slide on the server and return only the result.Usage
Arguments and flags
JSON inputs
See dataset query for the JSON structure of filters, metrics and sort. Chart queries do not use the join parameters.Examples
Aggregate the data of a chart
TerminalOutput and limits
The table queried is the chart’s table as published, with values the chart has already aggregated. Grand total and subtotal rows are not in the table; they come back in result.excluded_totals, so use them for overall values instead of adding up rows. Summing or averaging a column the chart already averaged, counted distinct or computed as a ratio carries the warning CHART_REAGGREGATION; then say so or use excluded_totals. When you report numbers, also say the chart’s aggregation, chart.filters, context.parameters and context.data_as_of. There is no paging. When has_more in the result is true, narrow the filters or aggregate.Troubleshooting
Related: expl server chart get-data, expl server chart get-image
expl server desktop mcp-setup-guide
Return the steps and links for connecting an AI tool to Exploratory Desktop.Usage
Arguments and flags
Examples
Get the steps
TerminalOutput and limits
It only returns steps and links. It does not install or run anything, and it does not operate Desktop. There is no paging.Troubleshooting
Related: Sign in
