This is a risky AI-agent setup CLI, but source inspection found no npm install-time execution, credential exfiltration, stealth payload, or concrete malicious chain. Its explicit-user-command control-surface mutations warrant a warning rather than a block.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
User runs the CLI, completes prompts, and confirms configuration.
Impact
Future Kimi sessions may load package-installed hooks with tool/MCP auto-approval; remote installer execution is supply-chain exposure.
Mechanism
AI-agent config/hook installation plus remote dependency bootstrap
Evidence
package.jsondist/index.jsdist/installer.jsdist/setup.jsdist/targets/hooks-export.jsdist/targets/kimi-local.jsdist/targets/sync.js$HOME/.kimi-code/config.toml$HOME/.grok/config.toml$HOME/.grok/trusted_folders.toml.kimi-code/local.toml.claude/hooks/