My name is Aziz Mohseny. I've built a personal tool called FFB (Font Feature Builder), and I've now reached a stage where technical feedback from this community would be really valuable to me.

The idea:
Instead of drawing a separate glyph for every dotted Arabic letter, we decompose the font into dotless skeletons and separate mark glyphs (dots and diacritics). At runtime, GSUB composes the skeleton and marks, and GPOS positions the marks from anchors.

The goal is to keep glyph counts manageable, let the designer focus only on skeletons, marks, and anchors, and allow them to develop the font without needing a font engineer.

What FFB currently does:

• Generates GSUB for isol / init / medi / fina

• Generates GPOS for mark and mkmk

• Generates curs from Ex / _Ex anchors

• Includes a browser-based tool (no install) for syncing anchor coordinates from a UFO

I've just published v0.1.0. The code is MIT-licensed, but the sample font glyphs are proprietary and for demonstration only.

In the next stage, I plan to add alternate glyph handling (features like calt and rclt), and also a tool for adjusting kerning of glyphs, dots, and marks. Ideally, I want the dots and marks to be fluid — all emerging from a single, shared fixed origin, while taking their best position relative to neighboring glyphs (contextual GPOS).

What I'm looking for:

• General feedback on the approach and the generated feature code

• Reports on any technical issues when compiling or testing

• Thoughts on whether this method aligns with how you actually build Arabic-script fonts

You can try the tool here:
https://ffb-flame.vercel.app

Repository and release notes:
https://github.com/AzizMohseny/FFB/releases/tag/v0.1.0

I know this is an early version and I'm not claiming it's production-ready. My goal is to understand where this approach breaks down and what you actually need.

Thanks for your time.
— Aziz Mohseny