The postinstall hook performs automatic setup of a first-party agent skill in detected agent homes, so this warrants a warning for agent extension lifecycle risk. The inspected source does not establish credential theft, broad agent settings changes, or another blockable attack.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
A supported npm or pnpm global install, or the matching pnpm add lifecycle path, triggers the hook.
Impact
The skill becomes available to agent hosts after package installation, which is an agent extension lifecycle risk.
Mechanism
The hook loads the package’s bundled gd-cli skill and syncs it to the canonical location and skills folders under detected agent roots.
Evidence
package.jsondist/bin/postinstall.jsdist/src/features/skill/bundled-skills.jsdist/src/features/skill/installer.js~/.agents/skills/gd-cli~/.claude/skills/gd-cli (when detected)~/.codex/skills/gd-cli (when detected)~/.cursor/skills/gd-cli (when detected)~/.config/opencode/skills/gd-cli (when detected)