Skill Testbot

Skills break in predictable ways: wrong CLI flags, nonexistent commands, steps that assume context that was never established, API endpoints that changed. The author cannot see these problems because they already know how it works.

Skill Testbot sees them.

What it checks:

• CLI commands and flags (do they actually exist and work as documented)
• API endpoints and request formats (are they correct)
• Logical flow (does each step follow from the last)
• Completeness (will a stranger be able to follow this without guessing)
• Config fields (do they actually exist in the tool being documented)

When to run it:

• Before publishing any skill to a marketplace
• Before installing a third-party skill you have not verified
• After updating a skill following an API or CLI change

How it works: It spawns an independent sub-agent that reads your skill file fresh — with no prior knowledge — and systematically checks every factual claim. You get a report with specific errors, not vague suggestions.

No dependencies required.