browser-act-skill-forge

SUSPICIOUS: the core purpose matches browser exploration and local skill generation, and install provenance is largely coherent and same-vendor. However, the skill normalizes bulk scraping/replication, contains dynamic `eval` verification, and instructs autonomous testing/install/execution with minimal approval, making its operational footprint broader and riskier than a typical documentation or helper skill.

No explicit malware behavior (e.g., credential theft, persistence, destructive actions, or specific exfiltration to known malicious destinations) is shown in this fragment. However, it contains a high-risk supply-chain primitive: eval of dynamically generated JavaScript produced by Python wrapper stdout via shell command substitution. This makes the overall system sensitive to parameter injection and upstream compromise. To assess real malicious intent, the actual scripts/*.py contents and the exact generated JS payloads/targets must be reviewed end-to-end.