diff --git a/content/docs/mcp_servers/index.mdx b/content/docs/mcp_servers/index.mdx
index 847cb162a..245575e2c 100644
--- a/content/docs/mcp_servers/index.mdx
+++ b/content/docs/mcp_servers/index.mdx
@@ -13,6 +13,9 @@ Use these guides to configure specific MCP servers with the right transport, OAu
   <Cards.Card title="Salesforce MCP" href="/docs/mcp_servers/salesforce" arrow>
     Configure Salesforce Hosted MCP servers with an External Client App and per-user OAuth.
   </Cards.Card>
+  <Cards.Card title="Local Commander MCP" href="/docs/mcp_servers/local_commander" arrow>
+    Let users work on files, Git, commands and local services on computers they approve.
+  </Cards.Card>
   <Cards.Card title="MCP overview" href="/docs/features/mcp" arrow>
     Learn how MCP works in LibreChat before adding product-specific servers.
   </Cards.Card>
diff --git a/content/docs/mcp_servers/local_commander.mdx b/content/docs/mcp_servers/local_commander.mdx
new file mode 100644
index 000000000..d44a1a25f
--- /dev/null
+++ b/content/docs/mcp_servers/local_commander.mdx
@@ -0,0 +1,97 @@
+---
+title: Local Commander MCP
+icon: Monitor
+description: Let LibreChat users work on computers they approve (files, Git, commands and local services) through Local Commander's hosted MCP server with per-user OAuth.
+---
+
+[Local Commander](https://localcmd.app) lets a model work on computers the user approves: read and
+edit files in approved folders, check Git, run commands and look at local services. It is a hosted
+MCP server at `https://localcmd.app/mcp` plus a small agent on each computer. Each LibreChat user
+signs in with their own Local Commander account and works only with the computers that account has
+approved.
+
+<Callout type="info" title="No client credentials needed">
+  Local Commander supports OAuth dynamic client registration. LibreChat registers itself on the
+  first sign-in, so there is no client ID or secret to create or store in `.env`.
+</Callout>
+
+## Prerequisites
+
+- A running LibreChat instance with `librechat.yaml` mounted or otherwise loaded.
+- A Local Commander account at [localcmd.app](https://localcmd.app) for each user who connects.
+- At least one computer with the Local Commander agent (Node.js 20 or later; Windows, macOS or
+  Linux).
+
+## Setup
+
+<Steps>
+  <Step>
+
+### Connect a computer
+
+On the computer the model should work on, run:
+
+```bash
+npx local-commander-agent@latest
+```
+
+Approve the pairing at [localcmd.app](https://localcmd.app). Repeat this on every computer you want
+to use.
+
+  </Step>
+  <Step>
+
+### Add Local Commander to `librechat.yaml`
+
+```yaml filename="librechat.yaml"
+mcpServers:
+  local-commander:
+    type: streamable-http
+    url: https://localcmd.app/mcp
+```
+
+Restart LibreChat after changing `librechat.yaml`.
+
+<Callout type="info" title="Strict MCP domain allowlists">
+  If your `librechat.yaml` also configures `mcpSettings.allowedDomains`, add `localcmd.app`.
+</Callout>
+
+  </Step>
+  <Step>
+
+### Sign in
+
+1. Open **MCP Settings** in the side panel. `local-commander` shows **Needs Auth**.
+2. Click **Connect**, then **Continue with OAuth**.
+3. Sign in to Local Commander if asked, check the permissions on the "LibreChat MCP Client" screen
+   and click **Allow connection**.
+
+LibreChat stores the tokens and loads the Local Commander tools. Finish the sign-in within 10
+minutes; LibreChat cancels an OAuth flow that takes longer, and you start again with **Connect**.
+
+  </Step>
+  <Step>
+
+### Use it in a chat
+
+Keep `local-commander` selected under the message box and ask, for example, "list my computers" or
+"run the tests in my project".
+
+  </Step>
+</Steps>
+
+## Troubleshooting
+
+- **The model provider could not complete this request:** the tool descriptions take about 18,000
+  tokens of context. With a local model, give it a context window of at least 32k tokens.
+- **The sign-in page says the connection could not be finished:** the OAuth flow timed out. Click
+  **Connect** in MCP Settings again and approve within 10 minutes.
+- **The model reports no computers:** start the agent on the computer and check that it shows as
+  online at localcmd.app.
+
+## What the model can do
+
+That depends on the access the user gave each computer: Full Access by default after approval, or
+only selected folders. Folder rules prevent accidents; they are not a sandbox. Users can see,
+cancel and revoke work at [localcmd.app](https://localcmd.app), and remove LibreChat's access there
+at any time.
diff --git a/content/docs/mcp_servers/meta.json b/content/docs/mcp_servers/meta.json
index 396b0e2c6..d9bd60b24 100644
--- a/content/docs/mcp_servers/meta.json
+++ b/content/docs/mcp_servers/meta.json
@@ -1,5 +1,5 @@
 {
   "title": "MCP Servers",
   "icon": "Blocks",
-  "pages": ["index", "google_workspace", "salesforce"]
+  "pages": ["index", "google_workspace", "salesforce", "local_commander"]
 }