FlagshipReady Live demo
Invoice & Receipt Extractor
Turns invoices, bills and receipts (PDF, scans, phone photos) into one checked row each, with line items, duplicate checks and a review list.
Claude skill + n8n workflow
$29
Get itAgents / Finance & ops / Claude skill packs
PDF to Sheet Extractor
Pulls tables out of PDFs into Excel and CSV, joins tables across pages, cleans numbers and checks Total rows.
Free Format: Claude skill#Finance & ops#Claude skill packs
Free
FreeReadyUpgrade to Pro: Invoice & Receipt Extractor · $29
Sold on Gumroad, which handles payment as the seller of record. Free download. Refund policy.
See it work
How PDF to Sheet Extractor moves through a job, step by step. The run below is an example output, written from what this agent does. It isn't a live run on your data.
Sample input
statement_oct.pdf (3 pages, one table split across pages 1-3)
Example output Illustration
tables.xlsx: 1 table, 86 rows (pages 1-3 joined) Numbers cleaned: "(1,200.00)" -> -1200.00 Total row check: PASS Check these pages: none
What you get
- Skill folders you add to Claude
- A Python script behind each skill, so numbers are calculated, not guessed
- Sample data and the test suite, so you can check it yourself
- Instant download from Gumroad ($0)
What it does
- Extracts ruled and borderless tables from text-based PDFs
- Joins tables that run across pages
- Cleans numbers like 1.234,56 and (1,200.00)
- Checks Total rows and running balances, and flags pages to check by eye
Tools and costs you'll need
- Claude (Code, claude.ai or API) with Agent Skills
- Python with pdfplumber + openpyxl (preinstalled in Claude's sandbox)
Third-party costs are paid to those providers, not to us.
Test report
Re-run by the Oarsman QA gate on 2026-11-01: 24/24 checks and 420/420 cells on synthetic PDFs, robustness 3/3. Real public PDFs (Census, two Federal Reserve releases, a sample invoice, password-protected, rotated, scanned): simple tables fine; dense statistical tables come out packed or fragmented and are now flagged on the page. Not tested: Claude's visual reading of real scans, PDFs over about 50 pages.
Known limits
- Dense statistical tables (two-line headers, dot leaders) still need manual work; those pages are flagged.
- Not tested: Claude's reading of real scans, or PDFs over about 50 pages.
Live demo
Demos use fictional businesses and data.
Flagship Upgrade to Pro
Want the full version? Invoice & Receipt Extractor
Turns invoices, bills and receipts (PDF, scans, phone photos) into one checked row each, with line items, duplicate checks and a review list. $29 on Gumroad.
What's AI here, and what isn't
This is an AI agent: it uses an AI model to read, sort, score or draft. It runs in your own accounts. Anything it would send outside your business waits for your approval unless you choose otherwise, and it never needs your passwords. Our AI disclosure.
Related agents
ProReady
Finance Analyst Skills Pack
Four Claude skills for month-end work: clean messy exports, reconcile bank vs books, explain budget variance and build a month-end pack.
4 Claude skills
$29
Get it
FlagshipReady Live demo
Claude Skill Packs Bundle
All three skill packs in one download: Finance Analyst, SEO + AI Visibility and E-commerce Weekly Report. $79 instead of $87.
Bundle of 3 Claude skill packs (13 skills)
$79
Get itQuestions about PDF to Sheet Extractor
What does PDF to Sheet Extractor do?
Pulls tables out of PDFs into Excel and CSV, joins tables across pages, cleans numbers and checks Total rows. Extracts ruled and borderless tables from text-based PDFs. Joins tables that run across pages. Cleans numbers like 1.234,56 and (1,200.00). Checks Total rows and running balances, and flags pages to check by eye.
What do I need to run PDF to Sheet Extractor?
You need: Claude (Code, claude.ai or API) with Agent Skills; Python with pdfplumber + openpyxl (preinstalled in Claude's sandbox). Third-party costs are paid to those providers, not to us.
Is PDF to Sheet Extractor tested?
Re-run by the Oarsman QA gate on 2026-11-01: 24/24 checks and 420/420 cells on synthetic PDFs, robustness 3/3. Real public PDFs (Census, two Federal Reserve releases, a sample invoice, password-protected, rotated, scanned): simple tables fine; dense statistical tables come out packed or fragmented and are now flagged on the page. Not tested: Claude's visual reading of real scans, PDFs over about 50 pages.
How do I get PDF to Sheet Extractor?
It's free on Gumroad: press "Get it" to download it.
Is the extraction done by AI?
No, a Python script does the extraction and the arithmetic checks. Claude runs it, checks the result against the PDF and reads scanned pages.
Does it work on scanned PDFs?
Claude reads scanned pages visually and passes its reading through the same cleaning and checks. A misread digit in a column with a total shows up as a FAIL; other columns are not checked.
What PDFs does it struggle with?
Dense statistical reports with two-line headers, dot leaders or section breaks. The current version (v1.0.2) flags those pages under 'Check these pages' so you know to check them by eye.
Back to all agentsMore in Finance & opsMore in Claude skill packs