Read a .twbx with Claude Code or Cursor.

What is the direct answer to this workbook question?

A .twbx file is a ZIP archive containing a .twb XML workbook and packaged data files or a .hyper/.tde extract. To let Claude Code or Cursor read it, unzip the .twbx so the agent can parse the .twb XML, or drop the .twbx into the xmetric workbook inspector to list sheets, formulas, and packaged data files locally.

Why can’t a coding agent read a raw .twbx as text?

If you ask Claude Code or Cursor to read sales_dashboard.twbx directly, the agent sees a binary archive. A .twbx (Tableau Packaged Workbook) bundles two different artifacts inside a standard ZIP container:

1. The .twb XML file: Contains all worksheet definitions, dashboard layouts, shelf placements, filter rules, and calculated field formulas.
2. The Data/ directory: Holds packaged .hyper or .tde data extract files, or embedded CSV and Excel sources.

Large .twb files contain verbose formatting XML; reading the raw file from line 1 fills the coding agent's context window with font colors and coordinate offsets before it ever reaches your formulas.

How do I unpack a .twbx in my terminal?

Have your coding agent unpack the archive into a temporary folder so it can inspect the XML structure without binary noise:

mkdir -p unpacked_twbx
unzip -q report.twbx -d unpacked_twbx
ls -lh unpacked_twbx

Once unpacked, tell Claude Code or Cursor to ignore <style>, <layout>, and <cards> blocks and target only <datasources> and <worksheets>.

How do I extract calculated fields with an agent script?

Give Claude Code or Cursor this short Python snippet to pull every calculated field name, caption, and formula out of the .twb XML in seconds:

import xml.etree.ElementTree as ET

tree = ET.parse("unpacked_twbx/report.twb")
root = tree.getroot()

for col in root.findall(".//datasource/column"):
    calc = col.find("calculation")
    if calc is not None and calc.attrib.get("formula"):
        name = col.attrib.get("caption") or col.attrib.get("name")
        formula = calc.attrib.get("formula")
        print(f"{name}: {formula}")

Filter for <column> tags that contain a <calculation> child to focus on calculation definitions rather than formatting XML.

How do I inspect and open the workbook in xmetric?

Extracting raw formulas from XML still leaves internal field IDs like [Calculation_9481029384] hard to trace across views and dependencies.

Drop the same .twbx file into the xmetric workbook inspector (/inspect/) in your browser:

• Lists data sources, sheets, dashboards, and calculated fields with internal IDs resolved to readable names, plus broken or unused calculations.
• Saves packaged data files locally in the browser with no server upload, so you can save or chart a CSV inside a .twbx.
• Runs Level of Detail and table calculations in the formula playground (/play/) on sample data and shows the SQL, so you can compare the logic against the queries your coding agent writes.
• Opens the workbook in the xmetric app (/app/), where you can keep building on its sheets with shelves and calculated fields.

The public xmetric MCP server is live at https://getxmetric.com/mcp with no authentication or API key. Anonymous initialize, tool discovery and widget-resource checks passed on 2026-10-29. ChatGPT and Claude host rendering and tool selection remain unverified; their directory listings remain coming. The local-file flow needs a host that renders the xmetric board; a coding client that only reads MCP text cannot open a local file in that board.

How do internal Tableau field IDs map to readable formulas?

In .twb XML, a calculated field that refers to another calculated field uses its internal bracketed ID rather than its display caption:

// Raw formula inside .twb XML
IIF([Calculation_104928374] > 0, [Sales] / [Calculation_104928374], NULL)

// Resolved human-readable formula shown in xmetric
IIF([Total Orders] > 0, [Sales] / [Total Orders], NULL)

-- Equivalent SQL for your agent's dbt or DuckDB query
CASE
  WHEN total_orders > 0 THEN sales / total_orders
  ELSE NULL
END AS revenue_per_order

What is live, and what is still coming?

The public xmetric MCP server is live at https://getxmetric.com/mcp with no authentication or API key. Anonymous initialize, tool discovery and widget-resource checks passed on 2026-10-29. ChatGPT and Claude host rendering and tool selection remain unverified; their directory listings remain coming. See the dated public protocol receipt and setup and evidence limits.

The public server is available; host operation and directory approval are separate checks.

The file board parses supported .pbix, .twbx/.twb, CSV, Excel and Parquet files in your browser. Choosing a file shares supported schema, source names and calculation definitions with xmetric and the chat host; requested result cells, filters and SQL or DAX evidence are shared when answering. The original file is not uploaded by this flow. Small aggregates and labels can still be sensitive. Privacy details.

What data does this worked example use?

The initial view uses authored synthetic data, not customer data or a captured agent session.

The example record lists its file, calculation or SQL and expected values. The live browser result is computed by xmetric’s existing engine or workbook inspector. Samples are excluded from own-data activation reporting.