Scenario 02: Creating Missing Evaluation Scenarios

User Prompt

"Create evaluation scenarios for the docker-containerization skill before publishing publicly."

Expected Behavior

1. Agent locates the skill directory at skills/infrastructure/docker-containerization/
2. Creates evaluation-scenarios/ directory if it doesn't exist
3. Analyzes the SKILL.md content to understand skill purpose and workflows
4. Generates 5-8 comprehensive scenario files (scenario-01.md through scenario-08.md)
5. Each scenario includes: user prompt, expected behavior, success criteria, failure conditions
6. Scenarios cover diverse use cases from the skill (basic, intermediate, edge cases)
7. Uses concrete examples with measurable outcomes

Success Criteria

• Agent creates evaluation-scenarios/ directory
• Agent generates minimum 5 scenario files (scenario-01.md to scenario-05.md)
• Each scenario file has all four required sections (prompt, behavior, success, failure)
• Scenarios are specific to the skill domain (Docker containerization)
• Success criteria are measurable (files created, commands run, outputs verified)
• Failure conditions clearly indicate when skill was not applied correctly
• Scenarios cover breadth of skill capabilities (basic → advanced)

Failure Conditions

• Agent creates fewer than 5 scenarios
• Agent uses generic scenarios not specific to Docker containerization
• Scenarios lack measurable success criteria (vague "agent does well")
• Missing any required section (prompt, behavior, success, failure)
• Scenarios test implementation details rather than skill application
• Agent copies scenarios from another skill without adaptation