taokynang
Analysis
This is a Vietnamese guide for creating Claude skills with a simple optional validator, and the provided artifacts show no credential access, network calls, persistence, or destructive behavior.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
Source: unknown; Homepage: none
The registry metadata does not provide an external source repository or homepage, so users have limited provenance context.
if __name__ == "__main__": ... folder_path = sys.argv[1] ... success = validate_skill(folder_path)
The package includes a command-line Python helper that can be manually run against a chosen skill folder.