The company says every skill and plugin uploaded to ClawHub now runs through AIG as part of that review. ClawScan runs AIG and NVIDIA’s SkillSpector independently; an AI judge considers both scanners’ findings alongside the uploaded files to produce a final assessment.
OpenClaw reports that the combined system matched 86.9% of labels and correctly classified 98.6% of malicious cases in an evaluation on a fixed 556-case subset of SkillTrustBench. The company also says it shares anonymized scanner disagreements and confirmed false positives with Tencent for regression tests and improvements. These figures describe the reported benchmark evaluation; they do not establish performance across all real-world uploads.