Seonglae ChoCWA, Context Window Architecture
After thinking and response, let agent to choose to keep think procedure into the context or not.
The real value of an AI application isn't just the LLM, but the system intelligently managing its context.
The 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.
Overall methods
- Keep-It-All – The Simple Approach
- Sliding Window – Focus on Recent Messages
- Summarization – Distill the Important Parts
- Retrieval-Based Memory – Smart Recall
- Agent-Based Context Management – Access Through Tools
Context Engineering Realized: Context Window Architecture
After working on this for some months now I would like to put it out there based on the considerable attention being put towards “context engineering”. I am proposing the Context Window Architecture (CWA) – a conceptual reference architecture to bring engineering discipline to LLM prompt construction. Would love for others to participate and provide feedback. A reference implementation where CWA is used in a real-world/pragmatic scenario could be great to tease out more regarding context engineering and if CWA is useful. Additionally I am no expert by far so feedback and collaboration would be awesome.
https://mrhillsman.com/posts/context-engineering-realized-context-window-architecture/
The New Skill in AI is Not Prompting, It's Context Engineering
Context Engineering is the new skill in AI. It is about providing the right information and tools, in the right format, at the right time.
https://www.philschmid.de/context-engineering
