Cover Letter Prompt Templates That Stay Specific

Cover letter prompts stay specific when the template forces specifics in: the full job post, an evidence bank of your real results, a tone slot, a banned-cliché list, a hard length cap, and a recruiter-skim check at the end. You leave with a six-slot master template, three variants (career change, referral, short email), a method for building the evidence bank once and reusing it, and a skim prompt that flags lines any applicant could have written. The templates work in ChatGPT, Claude, and Gemini. A sibling guide covers the basic draft-from-a-posting workflow. Soft path: https://promptmake.net/text turns your rough notes into a structured prompt.

Why cover letter prompts drift generic

Ask a model to "write a cover letter for this marketing manager job" and you get a letter that fits a thousand applicants. The model fills gaps with safe filler: passion for the mission, strong communication skills, eagerness to contribute. A hiring manager reads that letter and learns nothing about you.

You fix that in the template. Each slot in a good cover letter prompt removes one place the model can guess. The job post tells it what the employer wants. The evidence bank tells it what you have done. The tone slot sets voice. The cliché ban blocks the filler it reaches for first. The length cap forces choices, and the skim check catches what slipped through.

This page assumes you already know the basic loop of pasting a posting and asking for a draft. That loop lives in https://promptmake.net/blog/ai-prompts-for-cover-letters, which walks through a ChatGPT draft and honesty audit. Here you get reusable templates with slots you fill once per application.

The six slots of a specific cover letter prompt

A specific cover letter prompt has six labeled slots, and the labels matter. Models treat a labeled block as data to use, while loose text mixed into instructions gets blended or ignored. Label each block in capitals, paste your content under it, and put the instructions last so the model reads the facts before the task. Fill each slot fresh for each application, except the evidence bank and the cliché list, which you build once and reuse. If you skip a slot, the model fills it with a guess, and guesses sound like everyone else. The first five slots appear below, and the skim check gets its own section further down.

Slot 1: the full job post

Paste the whole posting, including the company blurb and the nice-to-have list. Then ask the model to extract the top three requirements before it writes anything. That extraction step keeps the letter aimed at what the employer asked for, not what the model assumes a marketing manager does.

Slot 2: the evidence bank

The evidence bank holds five to ten short entries about your real work, each with an action, a result, and a number or name where you have one. The prompt tells the model to use only these entries for claims. This slot does more than any other to keep the letter honest and specific.

Slot 3: the tone slot

Pick one tone and describe it in plain words: "warm and direct, like a note to a respected colleague" or "formal and concise for a law firm." Add one sentence you wrote yourself as a voice sample. The model matches rhythm better from a sample than from adjectives.

Slot 4: banned clichés

List the phrases you never want to see: "I am writing to express my interest," "passionate," "team player," "fast-paced environment," "hit the ground running," "perfect fit," "detail-oriented," "leverage." Tell the model to replace each with a concrete fact from the evidence bank. The ban list forces substitution, which pulls in specifics.

Slot 5: the length cap

Set a word cap and a paragraph count: "250 to 320 words, four paragraphs." A cap forces the model to pick the two strongest proofs instead of listing all ten. Say what each paragraph does: opening hook tied to the role, two proof paragraphs, a close with a clear ask.

Build your evidence bank once

The evidence bank is the slot you reuse across every application, so spend an hour on it before you write a single letter. Pull from your resume, performance reviews, project notes, and messages where someone thanked you. Write each entry as one or two sentences in plain words. Include the result and a number where you have one, and name the tool, client type, or team size when a number does not exist. Aim for eight to twelve entries so the model can choose the right two or three for each posting. Store the bank in a plain text file you can paste in seconds. Update it after each project so it stays current.

Entry format: action, result, proof

Use a fixed shape: "Action: rebuilt the onboarding email sequence. Result: trial-to-paid conversion rose from 9% to 13% in one quarter. Proof: owned copy and testing, worked with one designer." The fixed shape lets the model lift facts cleanly without inventing connective claims.

Numbers, names, and scope

When you lack a metric, give scope instead: team size, budget range, number of accounts, tools used. "Managed 14 client accounts in HubSpot" beats "managed client relationships." Skip confidential client names unless you have permission; "a regional credit union" works.

Map evidence to the posting

Before drafting, ask the model to match evidence entries to the three requirements it pulled from the job post, and to flag any requirement with no matching entry. A gap tells you to either add a real entry or address it in one honest line, such as a related skill you are building. Never let the model fill a gap with an invented achievement.

Paste-ready cover letter prompt templates

Copy a template, fill the brackets, and paste into ChatGPT, Claude, or Gemini. Keep the labels in capitals. The master template covers most applications. The three variants add a few lines to it for situations the master template handles poorly: a change of field, a warm introduction from someone inside the company, and portals with no cover letter upload. Save your filled master template in the same file as your evidence bank. For each new job you swap the JOB POST block, rerun the mapping step, and keep everything else as it was.

Master template

"JOB POST: [paste full posting]. EVIDENCE BANK: [paste 8 to 12 entries]. TONE: [tone description]. VOICE SAMPLE: [one sentence you wrote]. BANNED PHRASES: [your list]. LENGTH: 250 to 320 words, four paragraphs. TASK: First, list the top three requirements from the job post. Second, match each to one or two evidence entries and flag any gap. Third, write the letter: paragraph one names the role and one specific reason this company, paragraphs two and three each prove one requirement with evidence, paragraph four asks for a conversation. Use only facts from the evidence bank. Replace any banned phrase with a concrete fact."

Career changer variant

Add to the master template: "CONTEXT: I am moving from [old field] to [new field]. TASK ADDITION: In paragraph one, name the change in one plain sentence. In the proof paragraphs, pick evidence that transfers, and state the transfer: what the skill was, where I used it, and how it applies to this role. Do not apologize for the change."

Referral or internal move variant

Add: "REFERRAL: [name, their role, how you know them, and what they said about the team]. TASK ADDITION: Mention the referral in the first sentence with the person's name and role. Keep one proof paragraph instead of two, and use the freed space for why this team in particular."

Short email version

"Using the same JOB POST, EVIDENCE BANK, and BANNED PHRASES, write a 120 to 150 word email for when the application portal has no cover letter field. Subject line under eight words with the role title. One opening line, one proof with a number, one line asking for a call. No attachments mentioned unless I list them."

Run the recruiter-skim check

Recruiters and hiring managers skim. After the draft, run a second prompt in the same chat that simulates that skim and hunts for generic lines.

Skim prompt: "Act as a recruiter who reads this letter in 30 seconds. List the three facts you would remember. For each, name the job requirement it supports. Then quote any sentence that could appear in another applicant's letter without changes, and rewrite it with a fact from the evidence bank. Finally, confirm every claim appears in the evidence bank and flag any that do not."

If the recruiter remembers nothing concrete, your proof paragraphs lack numbers or names. Go back to the evidence bank, not the tone slot. Read the final letter out loud once. Cut any line you would not say in an interview.

Common mistakes with cover letter prompts

• Pasting a summary of the posting instead of the full text. The model loses the exact words the employer used.
• Skipping the evidence bank. The model invents plausible achievements to fill space.
• Using adjectives for tone with no voice sample. The result sounds like a template.
• No length cap. You get six paragraphs that list every skill you have.
• Accepting the first draft. The skim check catches filler the first pass always leaves.
• Reusing one letter for similar jobs. Change the job post slot each time and rerun the mapping step.
• Treating this as the drafting walkthrough. The step-by-step ChatGPT flow lives in ai-prompts-for-cover-letters; this page owns the template slots.

Model notes for cover letter prompts (September 2026)

Fast chat models such as GPT-5.5 Instant, Claude Sonnet 5, and Gemini 3.5 Flash handle the master template well for first drafts. Use a stronger model such as GPT-5.6 Sol, Claude Opus 5, or Gemini 3.1 Pro for the skim check and honesty audit, where catching an unsupported claim matters more than speed.

Paste personal details with care. Remove your home address, phone number, and references before you paste into a chat tool, and add them back in your own document. Check your account's data settings if you do not want chats used for training.

Build cover letter prompts with PromptMake /text

Paste your rough notes, such as "marketing manager job, 6 years in B2B email, want warm tone, keep it short," into https://promptmake.net/text. The tool returns a structured prompt with labeled sections you can merge with the six slots above. Then add your evidence bank and the job post and paste into your chat model.

PromptMake writes prompt text. It does not write the letter itself or submit applications. As of September 2026, guests get about three free generations per day per tool and registered free accounts about five.

FAQ

What are cover letter prompts?

Cover letter prompts are instructions you give a chat model such as ChatGPT, Claude, or Gemini to draft a cover letter. The strong ones include the job post, your real evidence, a tone, banned phrases, and a length cap. Weak ones ask for "a cover letter for this job" and get filler back. The slots decide the quality more than the model does.

How do I stop AI cover letters from sounding generic?

Feed the model an evidence bank of real results and tell it to use only those facts. Add a banned-phrase list so it cannot reach for stock lines. Include one sentence you wrote as a voice sample. Then run the recruiter-skim prompt to catch any line another applicant could have written.

What should go in an evidence bank?

Put eight to twelve short entries in it, each with an action, a result, and proof such as a number, tool, or team size. Pull them from your resume, reviews, and project notes. Write in plain words so the model can lift them cleanly. Update the bank after each project.

How long should an AI cover letter be?

Aim for 250 to 320 words in four paragraphs for a standard application. Use 120 to 150 words when you send it as an email or the portal has a short text box. Put the cap in the prompt, since models run long without one. A shorter letter forces you to pick your two best proofs.

Can I use one cover letter prompt for many jobs?

Reuse the template, the evidence bank, and the banned list, but change the job post slot each time. Rerun the step that extracts the top three requirements and matches evidence to them. That step changes the letter's focus from job to job. A letter sent unchanged to five employers reads that way.

Is it honest to use AI for a cover letter?

It is honest when every claim comes from your own evidence and you review the final text. The template enforces that by limiting the model to your evidence bank and asking it to flag unsupported claims. Read the letter out loud and cut anything you would not stand behind in an interview. Some employers ask about AI use, so answer truthfully if they do.

Which tool helps me build cover letter prompts for free?

https://promptmake.net/text turns rough notes into a structured prompt you can paste into ChatGPT, Claude, or Gemini. As of September 2026, guests get about three free generations per day and registered free accounts about five. You still supply the job post and evidence bank yourself. The chat model then writes the letter.