Why am I getting a 400 error when using YouTube Analytics MCP Server on Pipedream due to the 'ids' parameter not being included in the final query?

Help
1

This topic was automatically generated from Slack. You can find the original thread here.

I was so excited to find the MCP server that Pipedream offers but I having trouble with the YouTube Analytics MCP: YouTube (Analytics API) MCP Server | Pipedream

I have connected my account, see image, but when the model tries to use the tool I get a 400 error.

I cannot post the response here because it is VERY long so here is the pastebin: YT Analytics Pipedream MCP - Pastebin.com

It seems that the problem is this:

The crucial ids parameter, which was provided in the initial arguments ("ids": "channel==(CHANNEL ID)"), was not included in the final list of query parameters sent to the API (debug.config.params). The API received the start/end dates, the metrics, and the video filter, but it didn’t receive the mandatory ids parameter telling it which channel’s data to query.

2025-04-22 - 003.jpg
2025-04-22 - 003.jpg1000×502 96.9 KB

2

Can you do me a favor and let me know if you can use that tool (action) here in this web-based test app? Demo - Pipedream Connect

3

It’s the same tool, it just isn’t going through MCP or any LLM here

4

I’d like to see if it works for you there or if you run into the same issue

5

Same error:

{
  "exports": {
    "debug": {
      "status": 400,
      "statusText": "Bad Request",
      "headers": {
        "content-type": "application/json; charset=UTF-8",
        "vary": "Origin, X-Origin, Referer",
        "date": "Wed, 23 Apr 2025 04:20:50 GMT",
        "server": "ESF",
        "x-xss-protection": "0",
        "x-frame-options": "SAMEORIGIN",
        "x-content-type-options": "nosniff",
        "alt-svc": "h3=\":443\"; ma=2592000,h3-29=\":443\"; ma=2592000",
        "transfer-encoding": "chunked"

But when I send the request using the API and the HTTP request node in n8n, it works fine. I use the same bearer token and everything. Also, I love that interface, it lets me see all the options the LLM can use and helps me in prompting it. MCPs often don’t tell the user all the the MCP can do in detail, like listing all the dimensions and metrics (I know I can look at the API docs as well but it nice to see everything there in that simple user interface).

6

Yea totally, it can be tricky to troubleshoot this stuff with LLMs! Would you mind taking a screenshot of the configuration so I can check it out on my end? Feel free to DM it to me if you prefer.