Seonglae ChoThe top priority is to implement chat history increasing as a long-term memory proxy using vector-based hierarchy or file system-related hierarchy.
Specific implementation can be a chat history RAG or file system-based retrieval.
A system to manage context just as CPU cache architecture, including levels, and layers of understanding. This enables mapping tasks to code edit locations. This can be implemented as a subagent, memory folder, or hooks to implement this routing hierarchy.
Agent Context Managers
Fully automated background agent
- Agents can create and manage sessions autonomously
- Agents should only be limited by model intelligence, not by context or tool availability
- Owning custom tooling is far more powerful than off-the-shelf tools
- Fast sandbox + verifiable execution
- Background coding agent fully integrated into internal development workflow
Dynamic context discovery
Load only the necessary information on-demand, while managing the rest in files. This was actually a commonly used approach, but it has proven quite useful for overall AI agent workflows. Long tool outputs are saved to files, and agents can query them using
tail, read, grep, etc. when needed. Chat history can also be referenced as files after summarization. These are the kinds of needs that naturally arise with frequent use.
