Skill Details
Runtime
(claude-4-6-sonnet)
Example user message
[userMessage]: Here's my draft SKILL.md for a unit-test-generator skill. Is it ready to list? [pastes SKILL.md]
Example agent response
## Scorecard
| Dimension | Status | Note |
| Triggering | weak | Description says "helps with testing" — too vague;
won't reliably fire vs. abstain |
| Demonstrable output | pass | Produces AAA-pattern test files |
| Example honesty (refund risk) | FAIL | Example 1 shows full-coverage
suite; body only generates happy-path tests |
| Example diversity | weak | Both examples are the same function type |
| Setup completeness | pass | Install path + first run present |
| Scope | pass | One clear job |
## Highest-impact fix
Example honesty. Before: example shows edge-case + error-path tests. After:
trim to the happy-path tests the body actually generates, or extend the
body to match. The current gap is your top refund cause.
## Secondary fixes
1. Triggering: rewrite "when to use" with concrete phrasings + a negative.
2. Example diversity: make Example 2 a different language/function shape.
## Listing-ready rewrite
[Full corrected package follows...]
By downloading this skill, you agree to our terms of service
CLAUDE-4-6-SONNET
Build a Claude Skill that actually sells. Listed skills fail two ways: the agent never triggers them, or the demo doesn't match real behavior — causing refunds and bad reviews. This skill takes you from idea to a complete, marketplace-ready listing: a tight trigger description, a concrete demonstrable output, two honest examples matching real behavior, and setup instructions that prevent install failures. Output maps straight onto the listing fields you fill in here.
...more
Added 4 weeks ago
