• TRIAGE
  • ROOT CAUSE
  • RECOVERY
  • HUMAN

How to troubleshoot failed pipeline executions with your AI

Connect Matillion to your AI client to turn messy execution logs into clear error reports. Once connected, your agent handles the heavy lifting of log analysis.

Ask AI about this page

Short answer

Why did my Matillion pipeline fail?

Your AI pulls recent execution history and specific pipeline configurations to identify the exact point of failure. You get a direct explanation of the error and a list of likely causes. This breakdown shows you exactly where to look.

Where every issue lands.

Troubleshooting outcomes

The AI categorizes every failure into actionable buckets.

TRIAGE

Error identification

The AI scans recent runs to find the specific error message. You get a clear summary of what went wrong.

ROOT CAUSE

Logic gap detection

By looking at the pipeline structure, the AI finds mismatches in data types or missing components. It points to the specific step that broke.

RECOVERY

Fix suggestions

The AI suggests specific changes to your orchestration or transformation logic. You receive a list of things to check before rerunning.

HUMAN

Manual intervention

If the error is a credential issue or a network timeout, the AI flags it as a system problem. You handle the actual connection or permission fixes.

The workflow

What your AI does when a pipeline fails.

The AI moves from broad execution history to specific pipeline details to find the needle in the haystack.

  1. Scan recent runs

    The AI looks through your recent history to find failed jobs and their status codes.

    list_executions

  2. Inspect the job

    Once a failure is found, the AI pulls the full configuration of that specific pipeline.

    get_pipeline

  3. Analyze error logs

    The AI compares the error message against the pipeline structure to find the mismatch.

    list_executionsget_pipeline

  4. Summarize findings

    The AI presents a plain English explanation of the failure and what to do next.

    list_executions

Try it

Copy these to start.

Use these prompts to begin troubleshooting.

Starting points

These are opening lines. Swap out the specific pipeline names or execution IDs with your own.

Matillion Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_cvCobHBQ94OGYTB46AOSlQ9YTUsvJ3NLbXMZS4ld/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Matillion capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "matillion-cloud-data-integration-elt-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_cvCobHBQ94OGYTB46AOSlQ9YTUsvJ3NLbXMZS4ld/mcp"
    }
  }
}
Copy into chat04

  • Check the last five Matillion executions and tell me which ones failed.

  • Look at the execution with ID 12345 and explain why it stopped.

  • Analyze the 'Daily_Sales_Load' pipeline and find any configuration errors.

  • Why did my most recent pipeline run fail? Check the logs and summarize the error.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Matillion once, then ask.

Just enter your credentials once. Your connection is encrypted, and you can immediately start asking questions about your data jobs.

Connect Matillion to your AI

FAQ

How this task behaves

  • 01

    Can the AI fix the pipeline for me?

    No. The AI can only read the execution logs and the pipeline configuration to tell you what is wrong. You have to make the changes in Matillion yourself.

  • 02

    Can the AI trigger a new run?

    No. The current tools only allow the AI to read execution history and pipeline details. It cannot start or stop jobs.

  • 03

    Does the AI see all my data?

    No. The AI only sees the metadata, execution logs, and the structure of the pipelines. It does not access the actual data moving through your tables.

  • 04

    How does it know which pipeline failed?

    It uses the list_executions tool to see the status of recent jobs and then uses get_pipeline to look at the specific failed job.

  • 05

    Can I ask about a specific error code?

    Yes. If you provide a specific error code from your logs, the AI can look up the corresponding pipeline to find the source.

  • More questions about Matillion? The Connector page answers them. See everything the Matillion Connector can do

Connect Matillion to Claude, Cursor, ChatGPT & more