{
  "context": "Agent was asked to read a public web article and write a markdown summary to summary.md. The correct approach is to use defuddle parse with the --md flag rather than WebFetch, then write a summary based on the extracted content.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Used defuddle instead of WebFetch",
      "description": "The agent invoked defuddle parse <url> rather than WebFetch or curl to retrieve the page content.",
      "max_score": 40
    },
    {
      "name": "Used --md flag",
      "description": "The defuddle command included the --md flag to request markdown output.",
      "max_score": 25
    },
    {
      "name": "Produced summary.md",
      "description": "A file named summary.md exists in the working directory containing a readable summary.",
      "max_score": 20
    },
    {
      "name": "Summary contains meaningful content",
      "description": "summary.md contains structured markdown (headings or bullet points) with substantive content rather than just the raw page dump.",
      "max_score": 15
    }
  ]
}