Troubleshooting

New
  • Published on Mar 6, 2026
  • 1 minute(s) read
  • Ben F
 Prev Next

When things don't work as expected with Kentik’s AI Advisor MCP Server, check these common scenarios:

Authentication & Connection Failures

  • Symptoms: "Authentication failed", "Invalid credentials", or timeout errors.

  • Quick Fixes:

    • Verify Your Region: The most common error is mixing up the US and EU endpoints. Ensure your URL matches your account region (api.kentik.com vs. api.kentik.eu).

    • Check Your Token: Double-check that your API token is active in the Kentik Portal and hasn't been revoked.

    • Firewall Rules: Ensure your local network allows outbound HTTPS connections to the Kentik API endpoints.

    • License Check: Confirm with your admin that your specific account has AI Advisor access enabled.

MCP Server Not Appearing

  • Symptoms: The Kentik AI Advisor server doesn't show up in Claude Desktop or your chosen client.

  • Quick Fixes:

    • Validate Your JSON: A missing comma or mismatched quote in the claude_desktop_config.json is the #1 cause of this issue. Run your config through a free online JSON validator.

    • Check Your PATH: Ensure npx is installed and available in your system's PATH (specifically required for Claude Desktop).

    • Hard Restart: Completely quit (don't just minimize) and restart your client application to force it to re-read the configuration file.

AI is Ignoring Kentik Data

  • Symptoms: The AI responds to your network questions with generic advice, hallucinates answers, or fails to query Kentik.

  • Quick Fixes:

    • The "Kentik" Keyword: Ensure you are explicitly including the word "Kentik" in your prompt (e.g., "Show me my Kentik devices" instead of "Show me my devices").

    • Verify Connection State: Open your client's developer settings and verify the Kentik MCP server is actually showing as "Connected."

Slow Responses or Timeouts

  • Symptoms: Queries are taking a long time to return an answer or are timing out entirely.

  • Quick Fixes:

    • Patience for Complex Queries: Deep network queries can take 30-60 seconds to process. This is normal behavior. Watch the real-time progress indicators to ensure it's still working.

    • Rate Limiting: You may have hit the API rate limit. Wait 60 seconds and try again.

    • Prompt Strategy: Try breaking down massively complex questions into smaller, targeted follow-up questions using the ask_followup tool.

Related articles