BrandDocs is an open-source Claude Code and Codex skill bundle that learns a company's Word, PowerPoint or Excel template and generates new on-brand documents from it.

How is BrandDocs different from a generic AI document generator?

Generic AI document generators freely create documents and loosely imitate templates. BrandDocs extracts a Brand Profile from the original Office template and generates from the real template shell, so brand styles, structure and formulas are preserved by construction.

Which Office formats does BrandDocs support?

BrandDocs supports Word .docx, PowerPoint .pptx and Excel .xlsx through three agent skills: brand-docx, brand-pptx and brand-xlsx.

Does BrandDocs run locally?

Yes. BrandDocs is local-first and uses Python libraries such as python-docx, python-pptx, openpyxl, lxml and Pillow. Visual QA can optionally use LibreOffice, Poppler, PyMuPDF and Tesseract.

Local-first Claude Code & Codex skill

AI Office document generation from your real company templates.

BrandDocs is a set of agent skills that learn your Word, PowerPoint and Excel templates and generate faithful .docx, .pptx and .xlsx files from the original template shell, with brand drift blocked by verification.

View on GitHub Install Why it exists

Use cases

Generate branded Word reports

Start from a corporate .docx template and produce reports, memos or letters that keep the cover, table of contents, headings, callouts, lists, tables, native charts and diagrams in the template's structure.

Create PowerPoint decks from layouts

Extract real slide masters and layouts from a .pptx, then generate title, section and content slides from typed AI content, with native charts, diagrams and merged tables, without inventing off-brand layouts.

Fill Excel workbooks safely

Use named cells and regions in an .xlsx template while preserving formulas, workbook structure and the author's intended inputs.

How BrandDocs works

Extract a reusable Brand Profile (styles, layouts, structure and reusable component patterns) from a Word, PowerPoint or Excel template.
Generate new content from the original template shell using brand-agnostic typed blocks.
Verify styles, layouts, named ranges, formulas, residual template text and optional visual QA.

Why the output stays on brand

The generator never writes literal style names, colors or fonts. Those live in the extracted Brand Profile, and the QA gate fails when a profile points to an artifact that the template does not define.

Current status: alpha, with DOCX as the reference vertical.

BrandDocs vs generic AI document generation

Need	Generic AI document generator	BrandDocs
Use a real company template	Usually imitates the template visually.	Extracts the template into a reusable Brand Profile and generates from the original shell.
Preserve brand identity	Fonts, colors and structure can drift.	Uses only styles, layouts, ranges and artifacts the template actually defines.
Repeat the workflow	Requires re-explaining the brand each time.	Extract once, reuse the brand kit for future documents.
Excel formulas	Often treated as generic table content.	Named-region fills preserve shell formulas and workbook structure.

Install

Claude Code plugin:

/plugin marketplace add ferdinandobons/brand-docs
/plugin install brand-docs@brand-docs

Direct CLI

git clone https://github.com/ferdinandobons/brand-docs.git
cd brand-docs
python3 -m venv .venv
. .venv/bin/activate
pip install -r requirements.txt

FAQ

Is BrandDocs open source?

Yes. BrandDocs is MIT licensed and self-contained.

Does it work with Codex?

Yes. The skills are plain agent skills and the engine is a Python CLI.

Does it upload templates?

No. BrandDocs is local-first; brand-kit/ and generated outputs are git-ignored.