You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/docs/20-concepts/03-admin/05-mcp-servers.md
+45-2Lines changed: 45 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -110,7 +110,7 @@ In this case you would select `required` and `sensitive` options as well.
110
110
111
111
For multi-user setup, you would follow the same steps but would be configuring this to **SHARE** a common API key with ALL users.
112
112
113
-
### UVX: For Python-based packages
113
+
### UVX: For Python-based packages
114
114
115
115
If you found an MCP server like Duckduckgo and want it added to the Obot Gateway you would do the following.
116
116
@@ -148,6 +148,49 @@ You will need to select the container option from the drop down. Then provide th
148
148
149
149
You can also provide configuration through environment variables by filling in the configurations.
150
150
151
+
### Nanobot: For specialized STDIO deployments
152
+
153
+
The Nanobot runtime provides a specialized containerized environment optimized for MCP server deployments. This runtime is ideal for STDIO servers, where neither npx nor uvx runtimes suffice.
154
+
155
+
A few examples of when the Nanobot runtime is useful include:
156
+
157
+
- When your application needs some runtime dependencies that you can build into the image.
158
+
- When your NPM or Python package is not published or is published to a private registry, you can build the image yourself and publish it.
159
+
- When you have an STDIO MCP server that is written in a language that is not supported by the other runtimes (like Rust, Go, or Zig, for example).
160
+
161
+
Note that the `nanobot` binary must be included in the container image. You can install a prebuilt binary from the [GitHub releases page](https://github.com/nanobot-ai/nanobot/releases) or build it from source. An example Dockerfile for a building nanobot from source might be:
When configuring a Nanobot runtime, you need to specify:
187
+
188
+
-**Image**: The container image URI (e.g., `docker.io/myorg/nanobot-server:latest`)
189
+
-**Command**: The primary command to launch the MCP server (e.g., `my-custom-server`)
190
+
-**Arguments**: Optional command-line arguments to pass to the command
191
+
192
+
The Nanobot runtime handles starting the STDIO server and proxies requests via an HTTP transport to the STDIO server. Configuration parameters can be passed through environment variables using the standard configuration section.
193
+
151
194
## Configuration parameters
152
195
153
196
How configuration is handled depends on the server type:
@@ -197,4 +240,4 @@ After successfully adding a server:
197
240
- The server appears in the available servers list for authorized users
198
241
- Server entries can now be added to authorization groups for different teams
199
242
- Users can integrate the server into their clients to access tools in conversations and tasks
200
-
- Administrative monitoring of usage and auditing is available through the Admin panel
243
+
- Administrative monitoring of usage and auditing is available through the Admin panel
0 commit comments