This came up in Chinese AI dev circles this week: a set of seven Claude prompts for job hunting, shared with genuine enthusiasm. Most of them are fine. One of them is a trap.
The prompt in question is called the "Positioning Rewriter." You give Claude a target job title, and it rewrites your career summary and achievements to highlight relevant accomplishments and industry language. The post even includes a safety note: "Do not fabricate anything I haven't done."
That note is doing a lot of work. Maybe too much.
Here is what actually happens when you run a prompt like that. Claude does not know which parts of your resume are embellished already. It does not know whether the "led a cross-functional initiative" on line three was you coordinating two Slack threads or you genuinely owning a six-month project. It optimizes the language toward whatever sounds best for the target role, and it is very good at making ordinary things sound significant without technically lying. The result reads as confident, polished, and calibrated for keyword matching.
One comment in the thread put this more bluntly. Someone wrote that they had interviewed a candidate recently whose resume had clearly been AI-rewritten to the point of fraud. Halfway through the interview it became obvious the person could not back up what was on the page. The candidate disappeared from the call. That commenter suspected the network cable got pulled.
Now, that is one data point. But it points at something real: there is a gap between "the AI did not fabricate anything" and "the resume accurately represents what you can do on day one." Aggressive reframing fills that gap without crossing the technical line, and most people do not notice they are doing it.
The other six prompts in the set are genuinely useful in a way that sidesteps this problem. The "Hidden Role Finder" asks Claude to list job titles your experience already qualifies you for, including outside your current industry. The "Transferable Skills Translator" extracts what you are actually good at and maps it to sectors that value those things. The "Qualification Gap Analyzer" compares your resume to a specific job description and tells you honestly what you are missing. None of these require rewriting anything. They work with what exists.
Those prompts are doing something different from the Positioning Rewriter. They are helping you see your own experience more clearly, not helping you present it more persuasively. That distinction matters more than it might seem.
Job hunting advice has always contained a tension between honest self-presentation and strategic self-promotion. The usual guidance is to lead with your strongest true framing, use the vocabulary of the target field, and not undersell yourself. That is reasonable. The problem is that AI makes it much easier to slide from "better framing" to "framing that requires a version of you that does not exist yet" without realizing it happened.
The safety disclaimer in the original prompt does not actually prevent this. It tells Claude not to invent experiences. It does not tell Claude to flag when the framing is getting ahead of what the experience can support. Those are different requests, and only one of them is being made.
If you want to use AI for job search prep, the more honest version of the Positioning Rewriter looks like this: ask Claude to rewrite your summary for a target role, then ask it to identify every claim in that rewrite and cite the specific resume line that supports it. If it cannot find the line, the claim comes out. That extra step is what the original prompt is quietly skipping. The set of prompts is not bad overall. But the framing around the Positioning Rewriter, that it will help you "stand out" for roles you might not have considered, glosses over the risk that standing out and being able to follow through are two separate problems.
So here is what I actually want to know: when you have used AI to rewrite resume language, did you find it made you more honest about your experience or less?