Transparency in Context Window Construction for AI Assistants

Question

**Bug Description**
I wish there was more visibility into how the context windows are constructed under the hood. I've found myself wanting this for other "agentic" development tools like Cursor, Replit, etc. too. More insight can make a big difference in debugging prompts. It might even be really helpful to open-source the terminal client code so that folks can more easily customize the terminal experience to their idosyncratic workflows. It might also make it easier for folks to use the Claude