• TRIAGE
  • ROOT CAUSE
  • STATUS
  • HUMAN

How to debug subgraph failures with your AI

Connect Goldsky to your agent to turn raw indexing errors into actionable fixes. Your AI handles the log digging while you focus on the code.

Ask AI about this page

Short answer

Why is my subgraph failing?

Your agent checks the pipeline status and pulls error counts and execution logs to find the specific cause of the failure. This identifies whether the issue is a schema mismatch, a RPC timeout, or a logic error in your mapping code.

Where every issue lands.

Debugging outcomes

The AI categorizes failures into these specific buckets.

TRIAGE

Identify the failure type

The agent uses error counts to determine if the pipeline is completely stuck or just lagging. You get a clear sense of the problem's scale.

ROOT CAUSE

Pinpoint the exact error

By scanning execution logs, your agent finds the specific line or event causing the crash. You skip the manual grep through thousands of lines.

STATUS

Check pipeline health

The agent monitors the current state of your indexing. You'll know immediately if a restart or a deployment is required.

HUMAN

Manual code fixes

The AI identifies the error, but you still need to rewrite the mapping logic or update the schema. The agent hands you the context to do it.

The workflow

What your AI does when a pipeline fails.

Instead of you hunting through dashboards, your agent performs the heavy lifting of data retrieval.

  1. Check health

    The agent first looks at the current state of your indexing to see if it is active or halted.

    get_pipeline_status

  2. Quantify errors

    It pulls the total number of errors to see if this is a minor hiccup or a total failure.

    get_pipeline_error_count

  3. Scan logs

    The agent dives into the execution logs to find the specific error messages and stack traces.

    get_pipeline_logs

  4. Summarize cause

    It combines the status, error counts, and log data to tell you exactly what went wrong.

    get_pipeline_logs

Try it

Copy these to start.

Use these prompts to kick off a debugging session.

Starting points

Replace the pipeline names with your actual Goldsky identifiers.

Goldsky Connector

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

Connector linkhttps://edge.vinkius.com/vk_preview_vKwCn24rK9jiYRm7bJaactLmNUpPZSSNAnWrQ0l6/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 — Goldsky capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "goldsky-web3-data-indexing-subgraphs-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_vKwCn24rK9jiYRm7bJaactLmNUpPZSSNAnWrQ0l6/mcp"
    }
  }
}
Copy into chat04

  • Check the status of my pipeline 'main-subgraph-prod' and tell me if it's running.

  • How many errors have occurred in the 'test-indexing-pipeline' in the last hour?

  • Pull the latest logs for 'dev-subgraph-v2' and explain why it's failing.

  • Look at the error count for 'production-indexer' and summarize the most recent log errors.

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

Start here

Connect Goldsky once, then ask.

One connection gives your agent access to your indexing data. Your credentials stay encrypted and secure.

Connect Goldsky to your AI

FAQ

How this task behaves

  • 01

    Can the AI fix my subgraph code automatically?

    No. The AI reads the logs and identifies the error, but you must manually edit and deploy your code.

  • 02

    Can the AI restart a failed pipeline?

    No. The tools only allow the AI to read status, error counts, and logs. It cannot trigger restarts or deployments.

  • 03

    Does the AI see all my Goldsky data?

    No. It only accesses the specific pipeline data you authorize through the connection.

  • 04

    How often does the AI check for errors?

    The AI only checks when you ask it to. It does not monitor your pipelines in the background unless you set up an external automation.

  • 05

    Can the AI change my subgraph schema?

    No. The AI is limited to read-only access for debugging purposes.

  • More questions about Goldsky? See everything the Goldsky Connector can do

Connect Goldsky to Claude, Cursor, ChatGPT & more