Recipe Cookbook

🎯 What this is: A grab-bag of copy-paste starting prompts for common tasks. Find something close to what you need, paste it into Claude Code, and adjust the specifics. This is the page you'll come back to most.

How to use a recipe: the prompts use placeholders in [brackets] and assume your files are in ~/work-automation/inbox. Change those to match your real folders and field names — you know what your documents contain; tell Claude Code.

Two rules before every recipe

Work on copies, never your only copy of a file.
Ask for a dry-run / preview before anything gets moved, renamed, or overwritten — then check it before saying go.

A safe opener you can paste before any risky recipe:

Work only on copies in ~/work-automation/inbox, never touch originals elsewhere, and always show me a preview (dry-run) of what you'll change before doing it. When you rename or move files in bulk, also write a simple log of what changed so I can undo it. Don't connect to any online service — only the local files I point you at.

(And the umbrella rule: only run this on data your company's policy allows — locally downloaded files, nothing regulated unless you've checked. See Module 10.)

📄 PDFs

Extract fields into a spreadsheet

For every PDF in ~/work-automation/inbox, pull out [the client name, the date, and the reference number] and put them in an Excel file called summary.xlsx with one row per file plus a column for the source filename. First show me what you read from ONE pdf so we can confirm where those fields are.

Merge many PDFs into one

Combine all PDFs in ~/work-automation/inbox into a single packet.pdf, in alphabetical order, with a divider page before each showing its filename.

Split one PDF into many

Split [big-file.pdf] into one file per page, named page-01.pdf, page-02.pdf, and so on.

Read a scanned (image) PDF

These PDFs are scanned images — the text isn't selectable. Use OCR to read them, then [extract the fields above / give me the text].

Pull specific pages

From [report.pdf], save just pages [3 to 5] as a new file [excerpt.pdf].

Fill a PDF form from a spreadsheet

I have a blank form [form.pdf] and a spreadsheet [people.xlsx] with one person per row. Create one filled-in PDF per row, named after the person. Do one as a sample first so I can check the fields line up.

📊 Spreadsheets & Excel

Clean up a messy export

Open [export.xlsx] and describe its columns. Then: remove the columns [sys_id, internal_ref], rename [u_name] to [Name], format all dates as MM/DD/YYYY, and save as [export-clean.xlsx] — keep the original untouched.

Remove duplicate rows

In [data.xlsx], find rows that are duplicates based on the [email] column, show me how many there are, then save a de-duplicated copy.

Combine many spreadsheets into one

Combine every .xlsx in ~/work-automation/inbox into a single sheet, adding a column that records which file each row came from.

Compare two files

Compare [last-month.xlsx] and [this-month.xlsx] on the [member ID] column. Tell me which IDs are new, which disappeared, and which changed.

Summarize / count

From [tickets.xlsx], give me a summary: how many rows per [status], and the total per [category]. Put it on a new summary sheet.

Split one sheet into many files

Split [all-clients.xlsx] into one file per [region], each named after the region.

📁 Files & folders

Batch rename to a standard

The PDFs in ~/work-automation/inbox have inconsistent names. Each file's first page has [the client name and year]. Rename them to [LASTNAME_FIRSTNAME_YEAR].pdf. Do a DRY RUN first: a table of old name → new name, no changes yet. Also keep a log of the renames so I can reverse them.

Sort files into folders

Sort the files in ~/work-automation/inbox into subfolders by [file type / the month they were last modified]. Preview the plan first.

Find files by date

List every file in [folder] that was modified in the last [7] days, newest first.

Find duplicates

Find files in [folder] that are duplicates (same name or same contents). Just list them in a table — don't delete anything.

Bulk convert

Convert every [.xlsx] in ~/work-automation/inbox to [.csv], keeping the same names, into a new converted folder.

📝 Reports & documents

Generate a recurring summary report

Read all PDFs in a folder I specify, pull out [these fields], count how many are [pending vs. complete], and produce a one-page summary as both an Excel sheet and a readable PDF. Then tell me the single command to run it again next time.

Mail-merge style document generation

I have a template [letter.docx] with placeholders like {{name}} and {{date}}, and a spreadsheet [recipients.xlsx]. Generate one document per row with the placeholders filled in. Make one sample first.

Turn data into a readable write-up

From [results.xlsx], write a short plain-English summary of the key numbers and any notable changes, suitable for pasting into an email.

🔎 Data checks & quality

Find missing or invalid data

In [data.xlsx], flag every row that's missing [a name or a date], or where the [date] isn't a valid date. Give me just those rows so I can fix them.

Validate against a rule

Check [file.xlsx]: every [ID] should match the pattern [ABC-12345]. List any that don't.

Spot anomalies

Look at the [amount] column in [file.xlsx] and point out anything that looks off — blanks, negatives, or values far outside the normal range.

💡 When nothing here fits

You don't need a recipe for everything. Just describe your task plainly, lead with a dry-run, and let Claude Code propose an approach:

Here's what I'm trying to do: [describe it]. The files are in [folder]. Suggest a plan first, flag anything risky, and don't change anything until I say go.