deploy-uni-hook

SUSPICIOUS: the skill’s core purpose matches its blockchain capabilities, and its documented RPC/explorer flows are coherent, but it grants an AI agent high-impact real-world powers: generating arbitrary hook code, broadcasting on-chain transactions with a private key, writing deployment records on main, and chaining into a publication helper. This is not confirmed malware, but it is high security risk because autonomous crypto deployment and credential-backed broadcasting are inherently dangerous even with opt-ins and simulation gates.

The fragment is a deployment helper and shows no clear malicious supply-chain behavior or covert data theft. The principal security issue is that the deployer private key and potentially credential-bearing RPC URL are passed as command-line arguments despite comments claiming otherwise. Additionally, sourcing hook.env gives any party able to modify the staged project arbitrary shell execution. Restrict project and environment-file integrity, disable shell tracing and process exposure where possible, and use a safer secret-delivery mechanism such as a protected file descriptor or tool-supported environment/input method.