GuidesClaude Code

5 Claude Code jobs for people who don't code

By Stephen Thorn · 5 minute read ·

5 Claude Code jobs for people who don't code: the prompts

Claude Code is Claude working on your computer, so it can open, rename, move and change your actual files. These are the prompts for 5 file jobs, each one set up to work on a copy, show you the plan first, and list what it changed.

What you need

  • A Pro, Max, Team or Enterprise plan, or a Console account. Anthropic's setup page: "The free claude.ai plan does not include Claude Code access."

  • Claude Code installed. Anthropic's terminal guide for new users walks you through it, step by step: https://code.claude.com/docs/en/terminal-guide. That guide says: "You don't need to know how to code." If you'd rather skip the terminal, the Claude Code desktop app works without one.

  • One working folder for these jobs, for example a folder called claude-work on your Desktop.

  • For PDFs over 10 pages: a tool called poppler, which Claude Code uses to read long PDFs in sections. On a Mac with Homebrew, run brew install poppler. If it's missing, Claude Code tells you pdftoppm is not installed.

Before you start

  1. Put the files for the job in your working folder.

  2. Open a terminal, go to that folder and start Claude Code:

cd ~/Desktop/claude-work
claude

The docs: "By default, Claude has access to files in the directory where you launched it." For a folder outside it, such as Downloads, Claude Code may ask before it reads.

  1. Check your permission mode at the bottom of the screen. Depending on the mode, Claude Code asks before it changes files, or it doesn't. In Manual mode it asks before it edits a file or runs a command that isn't a read. In auto mode it doesn't ask: edits in your working folder go through, and a second model checks other actions. The PERMS guide shows how to check your mode and switch it.

  2. Want a hard stop before anything changes? Press Shift+Tab until the status bar says plan mode on. In plan mode, Claude reads your files and writes a plan, and edits stay blocked until you approve it.

The 5 prompts

Every prompt starts and ends the same way: make a copy and work only on it, show the plan and wait for your OK, then list what changed. Keep those lines in. Change the words in [brackets].

1. Rename a folder of files by date

In this folder, make a copy of the folder [folder name] called [folder name]-copy. Work only on the copy. Don't touch the original.

Rename every file in the copy so its name starts with a date in the format YYYY-MM-DD, then a short description, for example "2026-03-14 supplier invoice.pdf". Take the date from [the date written inside the file / the date in its current name]. If you can't find a date for a file, leave its name as it is and tell me.

Before you rename anything, show me a table of old name and new name for every file, and wait for my OK.

When you're done, list every file you renamed, old name and new name, and save that list as rename-log.csv in this folder.

2. Merge a pile of spreadsheets into one

This one is written for CSV files. If yours are Excel files, save each one as CSV first, or ask Claude how it would handle them before it starts.

In this folder, make a copy of the folder [folder name] called [folder name]-copy. Work only on the copy. Don't touch the original.

Merge every CSV file in the copy into one file called merged.csv. Add a column called source_file with the name of the file each row came from. Don't drop, change or reorder any values.

Before you merge, check that the files have the same columns. Show me your plan, the number of rows in each file, and any columns that don't line up and how you'd handle them. Wait for my OK.

When you're done, tell me the row count of merged.csv next to the total rows of the files you merged, and list every file you created or changed.

3. Pull the tables out of a PDF

In this folder, make a copy of [file name].pdf called [file name]-copy.pdf. Work only on the copy. Don't touch the original.

Find every table in the PDF. Before you extract anything, list the tables you found, with the page number and the column headings of each, and wait for my OK.

Then save each table as its own CSV file named after its page, for example table-page-4.csv. Keep every number exactly as it appears in the PDF. If a cell is hard to read, leave it empty and tell me which ones.

When you're done, list every file you created, and for each table, the number of rows in the PDF next to the number of rows in the CSV.

4. Sort your Downloads folder

Make a copy of my Downloads folder in this folder, called Downloads-copy. Work only on the copy. Don't touch the original.

Sort the files in the copy into subfolders by type: Documents, Spreadsheets, Images, Installers, Archives and Other. [Change the list to suit you.] Don't delete or rename anything, and leave any folders that are already there as they are.

Before you move anything, show me how many files would go in each subfolder, with 5 examples of each, and wait for my OK.

When you're done, list every file you moved, from and to, and save that list as sort-log.csv in Downloads-copy.

Check the sorted copy. When it looks right, you can swap it in for your Downloads folder yourself.

5. Build one summary sheet from all of it

In this folder, make a new folder called summary-copy and copy these files into it: [merged.csv, the CSV files from the PDF, and any other files to include]. Work only on the copies. Don't touch the originals.

Build one summary sheet called summary.csv from those files. I want one row per [customer / month / supplier / project], with these columns: [column 1], [column 2], [column 3]. Add a column that names the source file for each number.

Before you build it, show me the columns, where each one comes from, and how you'll handle a row that appears in more than one file. Wait for my OK.

When you're done, list every file you created or changed, and show me 3 rows I can check by hand against the source files.

The mistake to avoid

Running it on your only copy. Claude Code can rewind its own file edits with /rewind, but the docs say checkpointing "does not track files modified by Bash commands", and their examples are rm, mv and cp, the commands that delete, move and copy files. A rename or a sort done that way can't be undone with rewind. The same page calls checkpoints "not a replacement for version control". So keep the original where it is, let Claude work on the copy, check the result, and swap it in yourself.