The package has no unconsented install lifecycle behavior, so it does not meet the blocking threshold. Its explicit setup flow nevertheless creates high-privilege, persistent AI-agent configuration and installs third-party tooling.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
A user runs the CLI, completes its configuration prompts, and confirms setup.
Impact
The configured agents can receive broad shell, file, web, and MCP capabilities in the selected project.
Mechanism
Explicit setup installs dependencies and configures AI-agent extensions.
Evidence
dist/installer.jsdist/setup.jsdist/targets/kimi-local.jsstacks/_shared/skills/remove-ai-marks/SKILL.md.claude/settings.json.claude/hooks/.claude/skills/.kimi-code/local.toml.cursor/skills/ui-ux-pro-max/AGENTS.md
Network endpoints1
/Homebrew/install/HEAD/install.sh