4 Prompts to Decode Any Job Description in 3 Min

Most people read a job description left to right, nod along and send their resume. Then they're surprised when the screening call goes somewhere completely different. A job description isn't an HR wishlist, it's an encoded text. And decoding it manually takes time. That's exactly why it makes sense to build a prompt stack once and blast through any JD in minutes.

Why one prompt isn't enough

You can ask ChatGPT or Claude to "analyze this job description" and get a beautiful soup of everything at once. Honestly, they'll try to cover it all and end up saying nothing specific. The problem is that JD analysis is actually several distinct tasks. What the candidate REALLY needs to do. What the company is actually like. Where the red flags are hiding in the text. What to ask, and what to avoid asking. Each task needs its own focus. Short version: a stack of 4 separate prompts gives you far sharper results than one "analyze everything" request.

These 4 prompts work on ChatGPT, Claude, Gemini - any model with access to text. You paste the JD text after each prompt and read the response. Nothing complicated.

Prompt 1. Hidden requirements

Every job description contains two versions of requirements. The written ones - what HR copy-pasted from a template or what the recruiter asked to add "for scale". And the real ones - what the person genuinely can't be hired without. These are often different things. The first prompt extracts the real ones.

What to look for in the output: if the model returns more than 5 hidden requirements that don't appear in the Requirements section - the JD was written by someone who doesn't fully know who they're looking for. That's important information in itself.

Prompt 2. Red flags and company signals

Here's the nuance. Most people start looking for red flags only after the first interview, when something already feels off. But the job description often gives everything away upfront. There are specific phrases that correlate with toxic micromanagement, absent processes, or simply with a position that's been open three times and nobody stayed.

• "Competitive salary" with no range - either they don't know, or they know and are embarrassed
• "We are a family" - not always bad, but combined with "fast-paced" and no mention of PTO it gets interesting
• "Senior" title with ownership of everything and the budget - one candidate doing three jobs
• Zero mention of a team - either there is no team, or it's information they'd rather not share
• The LinkedIn posting has been open 60+ days unchanged - something is wrong with expectations or the process

Prompt 3. Decoding salary signals

Even when no range is listed, the job description contains plenty of indirect budget signals. The tech stack, company size, market type (B2B SaaS vs agency), number of requirements, equity mentions, seniority level vs actual responsibilities. All of this can be read. Honestly, this prompt is one of the most useful before a recruiter call - you go in with context, not guessing.

The result won't be accurate to the dollar - and that's fine. The goal isn't precision, it's entering the screening with your own frame so you don't undersell. And if you're already prepping for that call, Trackr's AI Coach can run through the full salary negotiation scenario in role-play mode - useful before the first contact.

Prompt 4. What to ask and what to skip

A recruiter screening is usually 20-30 minutes. Most of it goes to their questions. You get 3-5 questions, and they genuinely matter. Bad questions: "when will I hear back?", "what's the salary?", "how many vacation days?". Not because these don't matter, but because at the screening stage the recruiter often doesn't know the details themselves. Good questions show you were thinking about the role, not just firing off a templated application.

Actually, questions about how the team makes decisions or what the first month looks like are far more interesting to a recruiter than "what's your corporate culture?". The latter sounds like a template. The former sounds like someone who genuinely thought about the role.

How it works in practice

The scheme is simple. You find a job you're interested in. Before sending your resume, you run the 4 prompts - one at a time or all in separate messages. Read the output. Decide: is it worth pursuing, or are there signals to skip this one. If it's worth it - you already have a list of screening questions and an understanding of what to highlight in your resume.

1. 1Copy the full job description text, including the company description
2. 2Run Prompt 1 - hidden requirements, to understand what actually matters
3. 3Run Prompt 2 - red flags, to decide whether it's worth continuing at all
4. 4Run Prompt 3 - budget estimate, to enter the screening prepared
5. 5Run Prompt 4 - screening question list and topics to avoid
6. 6Log everything in your tracker so it doesn't get lost among 20 other applications

About the tracker - this matters. When you're processing more than 5-10 jobs at once, the analysis results genuinely disappear. Some people save them in notes, some in spreadsheets, but the most convenient setup is when everything is tied to the specific job. Trackr's 14-stage Kanban means each application moves through the pipeline and your notes stay with it. Not an ad, just genuinely useful when your application count grows.

A few more things worth knowing

• Models sometimes hallucinate about the company - if they "know" the company, verify the facts separately
• The more JD text you feed in, the more accurate the output. Don't cut the company description
• If the JD is short (under 200 words) - models struggle to extract anything useful, and that itself is a signal
• For foreign company JDs these same prompts work in English - just switch the prompt language
• If the company didn't list salary and refuses to share it before the technical interview - you have the right to not waste your time

And honestly, the most important result of this process isn't the question list or the salary estimate. It's that you enter the screening as someone who actually read the job description carefully. That comes through. Recruiters notice when a candidate understood the role. It gives you an edge before you've even opened your mouth.