Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Analyzes validation commands and test code in AI agent skills to identify patterns that fake quality signals. Detects exit code gaming, echo-only commands, tautological assertions, and commented-out tests. Rates each validation as SUBSTANTIVE, WEAK, or HOLLOW.
Manual code review can miss sophisticated test theater where real frameworks are invoked but assertions test trivially true conditions, which this tool flags through static analysis of assertion content.
When to use it
- Auditing agent marketplace skills before adding them to a production pipeline
- Reviewing community-submitted skills before merging a PR
- Comparing validation quality across a batch of published skills
- Flagging skills that claim test coverage but run no real assertions
- Deciding whether to trust a skill's published validation rating
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Check this Capsule JSON — its validation field claims 14 tests passed
- 1Parse the validation field from the provided Capsule JSON
- 2Break down each command in the validation chain
- 3Check each command for real assertion statements vs hardcoded success strings
- 4Inventory real assertions vs hollow outputs
- 5Generate a quality rating with evidence
HOLLOW rating: 0 real assertions found, 2 hollow outputs (hardcoded print statement claiming 14 tests, static echo). Recommendation to treat as unvalidated.