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

# LangSmith integration

> Connect LangSmith so ewake can read the traces and runs of your LLM applications during investigations.

> [!NOTE]
>
> **What you'll get:** ewake reads the traces, runs, threads, datasets and experiments of your LLM applications in LangSmith, so an investigation can see what an agent did step by step and what the model returned.

***

## How it works

The LangSmith card connects LangSmith's remote MCP server. Ewake calls only the LangSmith tools you select, with the permissions of the account you sign in with, or of the API key you give.

***

## Prerequisites

* A LangSmith account in the **US** or **EU** region. For a self-hosted LangSmith, use the [Custom MCP](https://docs.ewake.ai/integrations/custom/custom-mcp.md) card with the MCP URL of your instance.
* A LangSmith account with read access to the workspace, to sign in with, or a LangSmith API key
* Access to your ewake dashboard

***

## Configuration

**1. Open the LangSmith integration in ewake**

In your ewake dashboard, go to **Integrations** and click the **LangSmith** tile under **Add more**.

**2. Check the fields**

| Field | Value |
| - | - |
| **Name** | Filled with `langsmith`. A label to tell your LangSmith workspaces apart. Letters, digits, `-` or `_`. |
| **Region** | **US** (`api.smith.langchain.com`) or **EU** (`eu.api.smith.langchain.com`), the region of your LangSmith account |
| **Description** | Filled for you. Ewake reads it to decide when to consult LangSmith during an investigation. |

**3. Sign in to LangSmith**

Leave **Authentication** on **Sign in with OAuth** and click **Sign in**. LangSmith's sign-in page opens. LangSmith then acts as the account you sign in with, so sign in with read-only access.

Back in ewake, the panel says that you are signed in, and lists the LangSmith tools.

> [accordion] Connect with an API key instead
>
> Set **Authentication** to **No sign-in, or request headers (API key, bearer token)**. Under **Request headers**, add the header `X-Api-Key` with a LangSmith API key. Values are stored encrypted. Then click **Test connection** to list the tools.

**4. Choose the tools**

Under **Choose the tools Ewake can use**, click **Select read-only** to tick the tools whose names look read-only, then adjust by hand.

> [!WARNING]
>
> Ewake can only call the tools you select here, nothing else LangSmith offers, **even if it adds tools later**. Keep to read-only tools.

**5. Save**

Click **Save**.

> [!TIP]
>
> LangSmith is connected. Ewake consults it during investigations when the description matches what it's looking for.

***

## Connect more than one workspace

LangSmith is a **multi-instance** integration: connect as many workspaces as you need, each with its own name. Go to **Integrations** → **LangSmith** → **Configure**. The workspaces you already connected are listed under **Connected workspaces**, with the account each one signed in as. From there you can edit the tools of one, sign in again, or disconnect it. Add another under **Add workspaces**.

***

> [!NOTE]
>
> Ewake reads LangSmith with the permissions of the account you sign in with, or of the API key, and calls only the tools you select. Connect with read-only access, so that LangSmith refuses a write even if a write tool is selected.


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