Remote MCP (mcp.brightdata.com) returns empty results while showing "success" in event logs

[TweedBeetle]

Similar to #69, but affecting the remote MCP (mcp.brightdata.com/mcp?token=...) rather than the local npx version.

Symptoms

• MCP event logs show status: success for all scrape_as_markdown requests
• Tokens Used: 0 for all requests
• URL column is empty in event logs
• Requests never return content to the client (Claude Code in my case)
• Client times out waiting for response

Event Log Evidence

Date,Tool,Event,Client Name,URL,Status,Tokens Used
2026-02-02T11:10:14.175Z,scrape_as_markdown,POST /request,claude-code (via mcp-remote 0.1.36),,success,0
2026-02-02T11:09:25.290Z,scrape_as_markdown,POST /request,claude-code (via mcp-remote 0.1.36),,success,0
2026-02-02T11:06:48.398Z,scrape_as_markdown,POST /request,claude-code (via mcp-remote 0.1.36),,success,0

Note: All show success but URL is empty and 0 tokens used.

Configuration

{
  "command": "npx",
  "args": ["mcp-remote", "https://mcp.brightdata.com/mcp?token=<redacted>"]
}

Workaround

Switched to local npx MCP with a newly created Web Unlocker zone:

{
  "command": "npx",
  "args": ["@brightdata/mcp"],
  "env": {
    "API_TOKEN": "<token>",
    "WEB_UNLOCKER_ZONE": "web_unlocker_custom_for_mcp",
    "PRO_MODE": "true"
  }
}

This works, but users shouldn't need to abandon the remote MCP.

Expected Behavior

Remote MCP should either:

1. Return actual content (like local MCP does with fresh zone)
2. Return an error if the underlying zone is in a bad state (not silent "success")

Account

• Customer ID: hl_98bfdfae
• Email: christo@9592.tech
• Default zone mcp_unlocker was showing 1 request in dashboard with $0.0015 spent

Request

Could you investigate why the remote MCP's default mcp_unlocker zone returns empty results? The workaround is fine, but the silent failure mode (success + empty) is confusing.

[meirk-brd]

Hi @TweedBeetle ,

Thank you for reaching out.

Both the remote MCP and the STDIO version use the same underlying code path for scrape as markdown tool. I’ve also reviewed your account, and everything looks fine-no issues were detected.

I tried to reproduce the problem using the exact configuration you shared, and it worked as expected on my end. This suggests it may be an edge-case or intermittent bug, which I’ll dig into further to understand what might have caused it.

Could you please share which URLs you attempted to scrape when the error occurred? Did it happen with specific URLs, or with any request in general?

I also checked our internal logs, and they indicate that the data was returned successfully.

For reference, here are the results from my tests:

Are you still experiencing this issue, or does it appear to be resolved now?

If helpful, we can also jump on a quick call to walk through this together(I'll provide you with some credits as a thank you for helping us improve the MCP !). At the moment, we haven’t received similar reports from other customers.

Thanks again for flagging this, and I’m looking forward to your reply.