> ## Documentation Index
> Fetch the complete documentation index at: https://platform.atlan.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To act on Atlan objects, use the Atlan MCP server at https://api.atlan.com/mcp or the SDK. To install or publish skills from a terminal, use the `brain` CLI; `brain --map json` prints its command map. The `atlanai` CLI is deprecated. Run a read-only identity check before any write.
> The docs MCP server at /mcp searches these docs only. It cannot read or change Atlan objects.
> SDK packages: Python `atlanai` (PyPI) and TypeScript `@atlanai/sdk` (npm). Show Python first, then TypeScript.
> These docs cover Atlan Agent Registry, Atlan's hosted product. The open-source agent registry specification and its CLI (`brain`) are documented, released and maintained in their own repository; /open-source on this site links to it. Answer general specification or `brain` command questions from the project's own docs, and use these docs only for Atlan-specific setup such as the Atlan adapter, sign-in and workspaces.

# Find and read metrics

> Search the Metrics Studio catalog and read a metric's meaning, calculation, and release state so you know what agents get.

<Badge color="orange">Preview</Badge>

The Metrics Studio catalog lists every metric in a workspace and shows, for each one, whether agents can read it. Use it before you change a metric, when you need to know which calculation an agent is using, or when you are looking for a metric that may already exist. It complements the `/atlan-analytics` skill, which builds and looks up metrics through the Gateway MCP.

## Before you begin

Before you start, make sure you have:

* **Access**: Access to Metrics Studio and a workspace that has metrics. An empty workspace shows "No metrics in {workspace} yet" and a **Start your metric layer** panel instead.
* **Metric app**: An administrator has installed the Metric app in your account.

<Steps>
  <Step title="Open catalog">
    In the sidebar, under **Catalog**, choose **Metrics**. A count badge beside it shows how many metrics are in scope.

    Each row has a **Metric** column (the name, then a line with its warehouse connector, **Uses N**, **Used by N**, and its description), a **Release** column, an **Owner** column (**Unassigned** when none is recorded), and **Updated**. If two metrics share a display name, Studio shows the identifier under each name so you can tell them apart.
  </Step>

  <Step title="Search and narrow list">
    Type in the search box, whose placeholder is **Name or description**. Studio shows whether results came "from the gateway" (ranked) or matched on text only. Then narrow the list:

    * **Views**: **All**, **Unreleased changes** (the live release is behind the newest revision), **Not released**, and **Archived**.
    * **Filter**: by **Owner**, by **Warehouse**, and by **Replacement**. The **Replaced, still live** option appears only when a live metric has a replacement.
    * **Sort**: **Recently updated** (the default), **Name A–Z**, and **Relevance** (only after a ranked search).

    The footer reads "N of M shown", with the number of active filters. **Clear search and filters** resets the list.
  </Step>

  <Step title="Read release state">
    The **Release** column says what agents get today:

    | Label | What it means |
    | - | - |
    | **Live** with a version label | Agents get this version. The detail line names the channel, the revision it came from, and the release date. |
    | **Unreleased changes** | Newer revisions exist than the live version. Agents still get the live version. |
    | **Not released** | No release exists. Agents reading releases get nothing yet. |
    | **Archived** | Its releases are retired. |
    | **Unavailable** | Studio could not read the release history. |
    | **Replaced by {name}** | A replacement is named. Agents still get this metric until it is archived. |
    | **Pending review** or **Blocked** | The Registry is holding the newest revision. An authorized user can clear the hold; see [Release metrics](/studios/metrics/how-tos/release-metrics). |
  </Step>

  <Step title="Open metric">
    Choose the metric's name. Its page has these tabs:

    * **Overview**: a plain-words summary. **What it measures** lists what it includes and excludes, its granularity, unit, and parameters. **How it's calculated** shows the source table, joins, conditions, and aggregation as connected steps. **Joins and dimensions** appears when the metric declares dimensions. **Examples** shows the first four questions, with **View all N examples** for the rest.
    * **Definition**: the metric's parts, such as formula, source, dimensions, joins, parameters, and calculation settings.
    * **Examples**: each question with its SQL, and a **May be out of date** chip when its saved sample predates a change to the calculation.
    * **Dependencies**: what the metric uses and what uses it. A **Changed** badge means a dependency changed after the live version was released.
    * **History**: every saved revision. Every save creates a revision, and releasing a revision gives it a version number. Release tags are **Live**, **Released**, **Earlier**, **Pending**, and **Withdrawn**. **Earlier** is a superseded version that is kept and still viewable.

    If the revision you are reading is not the live one, a banner says "You are reading revision {n}, which is not released. Agents get {label}." Choose **See what agents get** to open the live version.
  </Step>

  <Step title="Verify what agents get">
    In **History**, open the version tagged **Live**. Its page, **Version {label}**, shows exactly what agents get now, and it does not change when newer revisions are saved until another version is released. It shows the formula, **What it measures in this version**, and **This version as JSON**. **Download release snapshot** saves the version as `{release_id}.json`.

    <Warning>
      A version page shows the stored definition. It does not show that the SQL runs or that the number is right, because Studio never executes SQL.
    </Warning>

    If the catalog shows "The catalog could not be read. Nothing below is a complete list.", the list is incomplete. Reload the page before you rely on it.
  </Step>
</Steps>

## Next steps

* [Retire metrics](/studios/metrics/how-tos/retire-metrics): Withdraw a release, name a replacement, or archive a metric.
* [Write metrics](/studios/metrics/how-tos/write-metrics): Revise a metric you just read.

## See also

* [Metrics Studio](/studios/metrics-studio): How revisions and releases relate.
* [MCP tool catalog](/tools/mcp/references/tool-catalog): The tools agents use to search metrics.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.