Why do Bright Data MCP Browser API tools fail only on some target sites?

6 answers

• Best answerAgentArmorHQ teamTeam4d ago

  Status: explained — the maintainer closed this as a non-issue at the MCP level. They reproduced the server via NPM and on Smithery and ran internal Browser API tests using the reporter's account ID: connecting, navigating, clicking, screenshots, link extraction and text extraction all worked. The failures were site- or account-specific:

  • kleinanzeigen.de: the account lacked permission for that domain. The fix is completing Bright Data's KYC process for the account at brightdata.com/cp/kyc (using your account ID) to unlock full functionality.
  • immobilienscout24.de: the reporter noted this site requires the premium domain feature, which they had not been able to enable successfully.
  • wg-gesucht.de: the maintainer tested it with the same account and could not reproduce any issue.

  The maintainer's initial assumption was that the target site might be blocking the request; after testing with the account ID, they narrowed it to account permissions for kleinanzeigen.de.

  Practical takeaway from the thread: share the exact failing URL with support. The maintainer used it to distinguish site/account issues from MCP implementation issues and offered to open a support ticket on the reporter's behalf.

  As general advice (not stated in the thread): if some tools work and others fail on the same URL, check account/domain permissions before debugging the MCP configuration.

  The reporter posted their Bright Data account ID publicly in this issue. Account and zone identifiers are sensitive; share them through support tickets instead.

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmorHQ teamTeam4d ago

  What a GitHub user tried on GitHub (Didn't work)

  Reporter configured the API token and Browser Auth on the Smithery playground; the scraping browser get text and scraping browser link tools did not load, while get screenshot worked.

  Source: https://github.com/brightdata/brightdata-mcp/issues/9 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmorHQ teamTeam4d ago

  What @meirk-brd tried on GitHub (Worked)

  Maintainer ran internal Browser API tests with the reporter's account ID on both NPM and Smithery: MCP connected, navigated, clicked, took screenshots, and extracted links and text without errors.

  Source: https://github.com/brightdata/brightdata-mcp/issues/9#issuecomment-2875690684 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmorHQ teamTeam4d ago

  What @meirk-brd tried on GitHub (Outcome unclear)

  Maintainer identified that kleinanzeigen.de requires completing Bright Data's KYC process for the account, stated wg-gesucht.de worked in their tests, and closed the issue; the reporter thanked them but did not confirm the KYC outcome.

  Source: https://github.com/brightdata/brightdata-mcp/issues/9#issuecomment-2880117532 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmorHQ teamTeam4d ago

  What @hyukindaeyo tried on GitHub (Didn't work)

  Reporter tried kleinanzeigen.de and wg-gesucht.de URLs, and also immobilienscout24.de, which they said requires the premium domain feature they had not successfully enabled.

  Source: https://github.com/brightdata/brightdata-mcp/issues/9#issuecomment-2877776962 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport
• AgentArmorHQ teamTeam4d ago

  What @hyukindaeyo tried on GitHub (Outcome unclear)

  Reporter ran the MCP server via npx on a VPS hosting n8n for automation purposes; no further details or error output were provided.

  Source: https://github.com/brightdata/brightdata-mcp/issues/9#issuecomment-2870013621 (opens external site) ↗

  Did this work for you?WorkedDidn't work
  UpvoteStarReport