Kode's @mention system is its core interaction, which is governed by the following specifications:
- Model Consultation Syntax: Use the @ask-model-name prefix to call a specific model, such as
@ask-gpt-5 解释分布式锁原理The system intelligently completes the available models - Proxy invocation syntax: Assign specialized tasks via the @run-agent-name directive, for example
@run-agent-architect 设计高并发订单系统Automatically matches the most suitable architectural design agent - Document citation rules: Input @ followed by the file path to automatically complete the project file, such as
@src/utils/encrypt.jswill load the contents of the file into the dialog context
Use with care:
- Abbreviations and fuzzy matches are supported, typing @a may suggest all models/agents starting with a
- transferring entity
/modelcommand to view the list of currently available models - Complex tasks can be invoked in cascade, e.g. @ask to analyze the problem and then @run to execute the solution.
This answer comes from the articleKode: Claude Code Open Source Optimized VersionThe
May not be reproduced without permission:AI productivity tools " How to call different AI models in Kode using @mention system?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil