Memory helps the Agent reuse your preferences and project experience in future tasks, reducing repeated context.
Proactive Memory
MiniMax Code can proactively identify information that will remain useful in later work, such as communication preferences, work habits, project conventions, and problem-solving experience. When related tasks appear later, the Agent can call on those memories as needed to reduce repeated explanations and corrections.What Belongs in Memory
- Communication preferences and output formats
- Project conventions
- Debugging experience and workflows
- Long-term information for a specific Agent or project
What Does Not Belong in Memory
- One-off task state
- Temporary logs
- Secrets
- Implementation details that are likely to expire
Memory is for long-term collaboration quality. It is not a replacement for project documentation or a place to store secrets.