parinshahAIbuilder

Review Markup — Development skill for Claude Code

Development community

Claude skill for marking review findings into the client's own files: colour coded PDF highlights with comments, Word tracked changes, and Excel cell marking with an index sheet.

How to install Review Markup

This entry records only its repository, not the path inside it, so there is no exact command to give. Open parinshahAIbuilder/review-markup and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Review Markup does

Claude skill for marking review findings into the client's own files: colour coded PDF highlights with comments, Word tracked changes, and Excel cell marking with an index sheet. Delivery layer for annual-report-review and ipo-offer-document-review.

Alternatives in Development

README

review-markup

A Claude skill for putting review findings into the client's own documents rather than writing them up separately: colour coded highlights with severity graded popup comments in PDF, tracked changes and margin comments in Word, and cell fills with comments and an index sheet in Excel.

It is the delivery layer for [annual-report-review](https://github.com/parinshahAIbuilder/annual-report-review) and [ipo-offer-document-review](https://github.com/parinshahAIbuilder/ipo-offer-document-review), and it stands on its own for any review that ends in a marked up file.

A review report is read once. A marked up document is read at the desk, next to the figure, by the person who has to change it.

What it does

python scripts/mark_pdf.py "statements.pdf" marks.json --dry-run
python scripts/mark_pdf.py "statements.pdf" marks.json

python scripts/merge_runs.py "audit report.docx" --out merged.docx
python scripts/redline_docx.py merged.docx edits.json
powershell -File scripts/verify_docx.ps1 "audit report - REDLINED.docx"

powershell -File scripts/mark_excel.ps1 "working papers.xlsx" marks.json -WhatIfOnly
powershell -File scripts/mark_excel.ps1 "working papers.xlsx" marks.json

Two colour palettes, kept apart because a reader who assumes severity when you meant status will read a closed item as critical:

Severity, for a first markup critical, high, medium, low
Status, for a later cycle verified as done, previously raised and still open, new issue

The status palette is the one clients ask for by the third cycle, usually in the words "highlight what is still not done".

The three rules it exists to enforce

**Do not damage the file.** `openpyxl` reads these workbooks correctly and writes them back missing parts. On a live three megabyte financial reporting set, a bare load and save lost **86 relationship parts**: chart XML, drawing relationships, printer settings, defined name scoping. The file still opened