[Skill Name]

[2-3 sentence description of what this skill accomplishes and when to use it.]

Prerequisites

Before running this skill:

1. Domain expertise engine installed (pip install -e lib/domain-expertise)
2. [Other prerequisites]

Input Required

• [Input 1]: [Description]
• [Input 2]: [Description] (optional)

Workflow

Step 1: Gather Information

[How to collect input from user or context]

Step 2: Load Domain Expertise

expertise query [domain] "[relevant query]"

This returns:

• Principles relevant to the task
• Rubric criteria for evaluation
• Contrast examples (WEAK vs STRONG)

Step 3: Execute Task

[Main logic - how to apply expertise to the input]

For evaluation tasks:

1. Compare input against rubric criteria
2. Score based on scoring guide
3. Identify specific issues (quote actual content)
4. Generate recommendations

Scoring Guide:

• 9-10: Excellent - [criteria]
• 7-8: Good - [criteria]
• 5-6: Adequate - [criteria]
• 3-4: Poor - [criteria]
• 1-2: Failing - [criteria]

Step 4: Output Results

Return results in this format:

{
  "input": "[what was analyzed]",
  "score": 0.0,
  "summary": "[1-2 sentence summary]",
  "strengths": [
    "[Specific strength]"
  ],
  "issues": [
    "[Specific issue with quote from input]"
  ],
  "recommendations": [
    "[OPTIMIZE] [Recommendation for existing element]",
    "[ADD NEW] [Recommendation requiring new content]"
  ]
}

Anti-Fabrication Checklist

Before returning results, verify:

• Every [OPTIMIZE] recommendation references something that exists
• Every [ADD NEW] recommendation is clearly flagged as new work
• Issues quote actual content, not paraphrased versions
• Recommendations don't reference features that don't exist

Example Usage

User: [example trigger phrase]

Claude: [brief example of skill execution and output]