{
  "$schema": "../schemas/rubric.schema.json",
  "evaluation_target": "description",
  "scale": {
    "min": 1,
    "max": 5
  },
  "reference_examples": {
    "judging_guidelines": [
      "Score only what is explicitly stated; do not infer capabilities or intentions.",
      "Penalize vague fluff, buzzwords, and over-claims.",
      "Do not reward verbosity; concise and clear beats long and padded.",
      "Evaluate trigger terms as: 'would a user naturally say this when they need this skill?'",
      "A missing 'Use when...' clause or equivalent explicit trigger guidance should cap completeness at 3.",
      "Descriptions must use third person voice ('Processes files', 'Analyzes data'). Penalize first person ('I can help') or second person ('You can use this') by reducing specificity score by 1.",
      "BOUNDARY GUIDANCE: Use odd scores (1, 3, 5) when the description clearly matches one anchor. Use even scores (2, 4) when it falls between two anchors. A score of 2 means noticeably below the midpoint; 4 means noticeably above it.",
      "ANTI-DRIFT: After assigning each score, re-read the anchors for the score above and below. If the description fits a neighboring anchor better, adjust the score to whichever anchor is the best fit. This prevents anchoring bias in either direction."
    ],
    "good_overall_examples": [
      "Extract text and tables from PDF files, fill forms, merge documents. Use when working with PDF files or when the user mentions PDFs, forms, or document extraction.",
      "Analyze Excel spreadsheets, create pivot tables, generate charts. Use when analyzing Excel files, spreadsheets, tabular data, or .xlsx files.",
      "Generate descriptive commit messages by analyzing git diffs. Use when the user asks for help writing commit messages or reviewing staged changes.",
      "Presentation creation, editing, and analysis. When Claude needs to work with presentations (.pptx files) for: (1) Creating new presentations, (2) Modifying or editing content, (3) Working with layouts, (4) Adding comments or speaker notes, or any other presentation tasks.",
      "Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply."
    ],
    "bad_overall_examples": [
      "Helps with documents",
      "Processes data",
      "Does stuff with files"
    ]
  },
  "dimensions": [
    {
      "id": "specificity",
      "name": "Specificity of Capabilities",
      "weight": 0.2,
      "question": "Does it describe concrete actions vs vague language?",
      "scores": [
        {
          "score": 1,
          "anchor": "Entirely vague; no concrete actions; pure abstract language",
          "example": "Helps with documents"
        },
        {
          "score": 2,
          "anchor": "Names the domain but actions are minimal or generic",
          "example": "Processes PDF files"
        },
        {
          "score": 3,
          "anchor": "Names domain and 1-2 concrete actions, but not comprehensive",
          "example": "Processes PDF files and extracts content"
        },
        {
          "score": 4,
          "anchor": "Lists several specific actions; minor gaps in coverage",
          "example": "Extracts text from PDF files, fills forms, converts pages to images"
        },
        {
          "score": 5,
          "anchor": "Lists multiple specific concrete actions; comprehensive coverage",
          "example": "Extract text and tables from PDF files, fill forms, merge documents, convert pages to images"
        }
      ]
    },
    {
      "id": "trigger_term_quality",
      "name": "Trigger Term Quality",
      "weight": 0.3,
      "question": "Does it include natural keywords users would actually say?",
      "scores": [
        {
          "score": 1,
          "anchor": "No natural keywords; only technical jargon or entirely generic language",
          "example": "Handles document object model manipulation"
        },
        {
          "score": 2,
          "anchor": "One or two generic keywords; missing the natural phrases users say",
          "example": "Works with files"
        },
        {
          "score": 3,
          "anchor": "Some relevant keywords but missing common variations or synonyms",
          "example": "Works with PDF files"
        },
        {
          "score": 4,
          "anchor": "Good keyword coverage; a few natural terms missing",
          "example": "PDF files, forms, document extraction"
        },
        {
          "score": 5,
          "anchor": "Comprehensive coverage of natural terms including synonyms and file extensions",
          "example": "PDF files, PDFs, forms, document extraction, .pdf"
        }
      ]
    },
    {
      "id": "completeness",
      "name": "Completeness",
      "weight": 0.35,
      "question": "Does it explicitly answer both 'what does this do' AND 'when should Claude use it'?",
      "scores": [
        {
          "score": 1,
          "anchor": "Missing both what and when, or both are extremely vague",
          "example": "Helps with documents"
        },
        {
          "score": 2,
          "anchor": "Has a vague 'what' and no 'when'; or only 'when' is present without 'what'",
          "example": "Use when working with documents"
        },
        {
          "score": 3,
          "anchor": "Has a clear 'what' but 'when' is missing or only weakly implied",
          "example": "Extract text and tables from PDF files, fill forms, merge documents"
        },
        {
          "score": 4,
          "anchor": "Has both 'what' and 'when'; 'when' could be more explicit or specific",
          "example": "Extract text from PDFs and fill forms. Use when working with PDF files."
        },
        {
          "score": 5,
          "anchor": "Clearly and explicitly answers both what AND when with concrete trigger phrases",
          "example": "Extract text and tables from PDF files, fill forms, merge documents. Use when working with PDF files or when the user mentions PDFs, forms, or document extraction."
        }
      ]
    },
    {
      "id": "distinctiveness_conflict_risk",
      "name": "Distinctiveness / Conflict Risk",
      "weight": 0.15,
      "question": "Is it clearly distinguishable from other skills; unlikely to trigger for the wrong skill?",
      "scores": [
        {
          "score": 1,
          "anchor": "Entirely generic; would conflict with virtually any skill",
          "example": "Helps with code and documents"
        },
        {
          "score": 2,
          "anchor": "Very broad; high overlap risk with many similar skills",
          "example": "Helps with document files"
        },
        {
          "score": 3,
          "anchor": "Somewhat specific but could still overlap with similar skills",
          "example": "Works with document files"
        },
        {
          "score": 4,
          "anchor": "Mostly distinct; minor overlap risk with closely related skills",
          "example": "Works with PDF and Word document files"
        },
        {
          "score": 5,
          "anchor": "Clear niche with distinct triggers; minimal conflict risk",
          "example": "Extract text and tables from PDF files, fill forms, merge documents. Use when working with PDF files or when the user mentions PDFs."
        }
      ]
    }
  ]
}