What this is
This is a simple framework using two dimensions — specificity and context completeness — to judge whether a prompt actually works. Specificity means whether the key decisions have been spelled out; context completeness means whether it's clear where the output will actually be used. Only when both dimensions are covered can Claude produce an answer that's close to what you wanted on the first try.
Many people only address one dimension and never think about the other at all — this is the most common source of prompt mismatch, and the core problem this framework addresses.
Why this exists
This framework exists because users often mistakenly assume 'writing more automatically means more accurate,' so when Claude's answer falls short, the instinctive reaction is to stretch the whole sentence longer — but if that extra length only adds specificity without ever supplying context, or the reverse, adding context while details stay vague, the answer still won't be accurate.
Breaking 'a good instruction' into two independent dimensions helps users precisely locate where the actual problem is, rather than vaguely lengthening the whole sentence without genuinely filling in the piece that was actually missing.
How this affects your decisions
This framework changes the order in which you check a prompt. You might previously have just written it and sent it, only going back to revise after getting an unsatisfying answer. Now you can self-check two things before sending: have I spelled out the key decisions, and have I made clear where this will actually be used. Confirming both before sending significantly reduces the number of back-and-forth revisions.
In practice, this means that beyond describing the task itself, you should add a line stating the usage context when writing a prompt — this habit costs little extra time but noticeably raises the odds of getting the result you wanted on the first try.
Advanced applications
Advanced users can flip this framework into a diagnostic tool for checking what's off in Claude's answer. If the answer's direction is right but the details are imprecise, that usually signals insufficient specificity — the fix is spelling out clearer key decisions. If the details are precise but the setting or tone is wrong, that usually signals insufficient context completeness — the fix is adding usage context. This diagnostic approach quickly pinpoints which dimension to fix, instead of rewriting the whole sentence by feel.
Another advanced move is fixing commonly used context information into Claude Projects' custom instructions — something like 'my output is usually a formal document for external clients, tone should be formal.' This way, the context completeness layer gets set once and applies long-term, and future prompts only need to focus on the specificity dimension, cutting down the burden of restating context every single time.
When people feel their communication with Claude isn't working, the first instinct is usually 'I need to write more detail,' stretching one sentence into a long paragraph — and the answer that comes back is often even further off. The problem usually isn't a lack of length; it's whether two independent dimensions are both being addressed: how specific the sentence is, and how clearly the usage context has been stated. Neither dimension is hard on its own, but many people only address one and never think about the other at all.
Specificity means whether you've described the task to the point where Claude doesn't need to guess the details itself. 'Write me a letter' is abstract — Claude doesn't know the length, tone, or whether a particular point needs mentioning. 'Write me a letter under 150 words, formal tone, reminding the client of the payment deadline' is far more specific — every key decision has already been made for Claude, so there's nothing left to guess. Specificity isn't about word count; it's about whether the key decisions have been spelled out.
Context completeness means whether Claude knows the eventual purpose and setting of the output. Both are 'write a letter,' but one attached to a formal contract and one sent to a close colleague need completely different wording. Both are 'draft a presentation outline,' but one for investors and one for an internal team need to emphasize different points. Many people get very specific about the task itself, but never mention where the output will actually be used, leaving Claude to fill that gap with the most conservative, generic assumption — which often ends up different from what was actually wanted.
Specificity without context completeness: you describe the task in fine detail, but Claude doesn't know where this thing ends up being used — the output might be precise but wrong for the setting. Context completeness without specificity: you mention 'this is for the client,' but don't spell out exactly what needs to be included, leaving Claude to guess the details, and the guess isn't necessarily what you wanted. A genuinely good instruction gets both dimensions right at once — clearly describing what needs to be done, and clearly describing where the result will actually be used.
Next time Claude's answer feels a bit off, don't rush to stretch the whole sentence longer. Instead, check these two dimensions separately: have I spelled out the key decisions (specificity), and have I made clear where this will actually be used (context completeness). The gap usually comes from one dimension being completely overlooked, and filling that in is often far more effective than simply writing a longer sentence.