Claude Prompts for Resume Tailoring (With a Reusable Project)

Claude's most useful feature for a job search is not a prompt, it is Projects: a workspace that keeps your master resume and your rules in one place, so each new application is a single message with the posting pasted in. Projects are available on Claude's free plan. Below is the setup, the project instructions to paste, and five prompts for the individual steps.

#What Claude documents about files, projects and limits

All figures come from Anthropic's help center and pricing page, as of October 2026.

The pricing page also says Pro gives at least five times the Free plan's usage per five-hour session. If you tailor one or two resumes a day, the free plan is normally enough.

#Set up a resume project once

1. In Claude, create a new project and name it after the role family, for example "Data analyst applications".
2. Upload your master resume to the project's knowledge. A text-based PDF or DOCX works; Anthropic notes that project files are read by text extraction.
3. Paste the project instructions below into the project's instructions field.
4. Start every application as a new chat inside the project, so earlier postings don't mix into the next one.

The master resume should be your longest honest version, with every role, tool and number you can defend. The assistant cuts from it; it should never have to add to it.

#Project instructions to paste

You help me tailor my resume (in project knowledge) to one job posting
per chat.
 
Hard rules:
- Never add an employer, job title, date, degree, certification, tool
  or number that is not in my resume.
- Use the posting's exact wording only where my resume already shows
  the same skill or responsibility.
- If a posting term would help but my resume does not prove it, write
  [confirm: term] in the bullet instead of claiming it.
- If a bullet needs a number I have not given, write [add: metric].
- Keep bullets to one line where possible, starting with a verb.
- End every answer with a list of what you changed and why.
 
Output format: plain text with standard headings (Summary, Experience,
Skills, Education). No tables, columns or icons.

The output-format line matters. Tables and columns can scramble when an applicant tracking system parses the file; our guide to resume parsers explains why.

#The one-message prompt for each application

Inside the project, open a new chat and paste:

Here is the posting for [job title] at [company].
 
1. List its required and preferred qualifications in its exact words.
2. Map each one to the line in my resume that proves it, or "none".
3. Rewrite my Summary, the bullets for my two most recent roles,
   and my Skills line for this posting, following the project rules.
4. List every change.
 
[paste the full posting]

#Five prompts for finer control

Use these in the same chat when you want to work step by step.

Requirement extraction

Quote every tool, certification, and responsibility in this posting.
Mark which are required and which are preferred. No paraphrasing.

Gap map

For each requirement, quote the resume line that proves it. Where
there is none, say "gap". Do not suggest wording yet.

Bullet rewrite

Rewrite only the bullets that match a requirement. Keep my numbers.
Put the posting's term in the first half of the bullet. Follow the
[confirm: …] and [add: …] rules.

Summary

Two sentences, using the posting's job title once, my years of
experience as stated in my resume, and my strongest number.

Change audit

List every claim in the new version that is not word-for-word
supported by my original resume, and quote the original line each
one is based on.

The audit is the step to keep even when you are in a hurry. A rewrite that reads well can still overstate scope ("led" where you "supported"), and that is the kind of change that surfaces in an interview.

#Check the output against what postings ask for

Your field's common terms are a useful sanity check on any rewrite. In the ATS Resume AI analysis of all 503 S&P 500 careers sites and 9,260 live job postings, June 2026 (study):

These describe the S&P 500 sample, not every employer. If a rewrite removed one of these terms from a bullet that genuinely covers it, put it back. For bullet structure, see ATS-friendly resume bullet points.

#Privacy choices before you upload

Anthropic's privacy center says it uses consumer chats to improve its models if you choose to allow it, and you can change this under Settings → Privacy with the Help improve our AI models toggle. If you allow it, de-identified data may be kept for up to 5 years; deleted chats leave back-end storage within 30 days.

Incognito chats are not used for training and are kept for 30 days, but they are not available inside projects. So the trade-off is: a project for convenience, or an incognito chat for a one-off tailoring session. Either way, remove your street address and phone number from the copy you give Claude; you can add them back in the final file.

Prefer a tool that applies these rules for you? ATS Resume AI does the same requirement matching and marks unprovable terms as [confirm: …] and missing numbers as [add: …]; it is free for 3 days with 5 AI rewrites and no card.

#Sources

• Anthropic, Upload files to Claude
• Anthropic, What are projects?
• Anthropic, Claude pricing
• Anthropic Privacy Center, Is my data used for model training?
• Anthropic Privacy Center, How do I change my model improvement privacy settings?
• Anthropic Privacy Center, How long do you store my data?
• Anthropic, Use incognito chats