kya

SUSPICIOUS. The skill’s capabilities broadly match its stated KYA/AWP identity and staking purpose, and its network targets appear purpose-aligned, but it relies on a downloaded prebuilt binary plus a curl-piped installer, forwards sensitive wallet/session material through external tooling, enables consequential on-chain/relayed actions, and introduces transitive trust via `awp-skill`. This is not clearly malicious, but it carries medium-to-high security risk for an AI agent skill.

The wrapper is primarily a CI bootstrap tool, but it introduces meaningful supply-chain and execution risks: it downloads and executes a remote `install.sh` from a fixed URL without integrity verification, and it prepends a potentially user-writable directory to PATH, which can enable `kya-agent` binary hijacking if the filesystem is compromised or shared. No direct malicious payload is evident in this wrapper alone, but its runtime installation behavior should be treated as high-risk and reviewed/pinned (e.g., hash/signature, pinned commit/artifact) and hardened (e.g., avoid broad PATH precedence, validate expected binary provenance).