Bug: Missing error details in n8n_executions response for failed workflows

[McBenac]

Problem

When a workflow execution fails, the n8n_executions tool with action: "get" returns execution metadata but not the actual error message from the failed node.

This makes debugging impossible for AI assistants/agents – they cannot see what actually went wrong.

Example

A workflow fails because the Code node uses require('crypto') which is blocked by n8n's security settings.

What the AI sees via MCP:

{
  "id": "391",
  "status": "error",
  "stoppedAt": "2025-12-09T20:06:03.977Z",
  "nodes": {
    "Process & Aggregate": {
      "status": "error"
    }
  }
}

What the n8n UI shows (the actual error):

Module 'crypto' is disallowed [line 2]

The AI has no way to know what went wrong and ends up in frustrating guess-and-fix loops.

Proposed Solution

When fetching execution details with mode: "full", include the error message from failed nodes:

{
  "id": "391",
  "status": "error",
  "nodes": {
    "Process & Aggregate": {
      "status": "error",
      "error": {
        "message": "Module 'crypto' is disallowed [line 2]",
        "description": "Error: Module 'crypto' is disallowed at..."
      }
    }
  }
}

Impact

This would significantly improve AI assistants' ability to debug and fix workflows autonomously, reducing the need for users to manually check the n8n UI.

Environment

• n8n-mcp version: latest
• n8n version: 1.121.1
• Client: Claude Code