Kode has designed a revolutionary @mention interaction system that enables precise invocation of models, agents, and project files through a unified syntax. The system contains three types of core functions: @ask-model-name for expert consultation, @run-agent-agent-name for task delegation, and @file-path for contextual reference. Together with the intelligent complementation technology, the system will prompt the available resources in real time after the user types @, and supports fuzzy matching and abbreviation input.
This design combines natural language instructions with structured operations. For example, the command "@run-agent-test-writer为@src/utils/date.js生成测试" not only maintains the ease of use of natural language, but also precisely specifies the executor and the object of the operation. In complex projects, the efficiency of this interaction is more than 40% higher than traditional command line tools.
This answer comes from the articleKode: Claude Code Open Source Optimized VersionThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil