Context Directness Spectrum
Classify context by how directly it comes from the user — their own words, indirect nearby sources, or boilerplate that only shapes the response.
Context gathered during context retrieval varies in how directly it comes from the user:
- Most direct — straight from the user: typed text in a help box, or the code block a user is currently editing.
- Indirect — from relevant nearby sources: searched documentation excerpts for a support request, or other open files/tabs that often contain relevant snippets for a coding assistant.
- Least direct — boilerplate text that shapes the response without coming from the user's specific situation at all, e.g. a top-of-prompt statement like "This is an IT support request. We do whatever it takes to help users solve their problems." Boilerplate introduces the general problem up front and later acts as glue connecting the more direct context fragments into a coherent document.
This directness axis is orthogonal to prompt element importance — the least direct boilerplate is often also the highest-priority element, since it frames how everything else should be read even though it says nothing specific about the user's actual request.
Sources
- Prompt Engineering for LLMs (John Berryman), ch. 4