gnodet commented on PR #23346:
URL: https://github.com/apache/camel/pull/23346#issuecomment-4502584037
_Claude Code on behalf of Guillaume Nodet_
Good question. The current STDIO approach requires putting the route
filename in the MCP config, which isn't user-friendly — you'd have to update
the config and restart your agent every time the route changes.
The plan is to take a different approach: a **JBang command** `jbang camel
mcp` that acts as a STDIO MCP server connecting to the dev console HTTP
endpoint of a running Camel app. The MCP config becomes simply:
```json
{ "command": "jbang", "args": ["camel", "mcp"] }
```
No URL, no route name, no port — ever. The command auto-discovers the
running Camel instance (JBang already has all the process discovery logic). If
multiple apps are running, you can pass `--pid=<pid>` or `--name=<name>`, but
the default just picks the single running instance.
This way:
1. Start your Camel app however you want (`jbang camel run myroute.yaml`)
2. The AI assistant spawns `jbang camel mcp` and talks to it over
stdin/stdout
3. The MCP command discovers the running app and translates MCP tool calls
to dev console HTTP calls
4. Change routes, restart the app — the MCP config never changes
The embedded Java MCP server stays useful for SSE transport later (native
MCP endpoint on the HTTP server), but for the AI assistant use case, this is
much cleaner.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
