I just published Glyphs MCP v1.0.8 and I’d love a few testers from the community to try it and report back (especially across different macOS + Glyphs 3 + Python setups).
What is Glyphs MCP?
It’s a Model Context Protocol (MCP) server for Glyphs that exposes font- and glyph-level operations as tools for AI/LLM agents (kerning, spacing review, glyph inspection/edits, etc.).
New in this release: a macOS Installer app (recommended)
To make setup easier, there’s now a signed + notarized macOS installer app (drag‑and‑drop DMG) that guides you through:
installing/updating Glyphs MCP.glyphsPlugin into ~/Library/Application Support/Glyphs 3/Plugins/
installing Python deps (Glyphs Python or a custom Python)
optionally configuring MCP clients (Codex, Claude Desktop, Claude Code, Antigravity)
A big ask: please help test the installer
Testing an installer thoroughly is genuinely hard as a solo developer—I can’t realistically cover every macOS/Glyphs/Python combination, upgrade path, permissions edge case, or corporate/locked-down machine scenario on my own. Even small feedback (“worked great on my setup” or “I hit this one dialog/error”) is valuable: it helps me fix issues faster and also helps other users by documenting what works (and what doesn’t) on real-world setups.
What I’m looking to test
macOS 13/14/15 (and Apple Silicon vs Intel if possible)
Glyphs 3.x
Python via Glyphs Python plugin vs python.org installs (3.11–3.13)
First install vs upgrade from an older version
Any Gatekeeper / permissions / “couldn’t load plugin” issues
If you try it, please reply with:
macOS version, Glyphs version, CPU (Intel/Apple Silicon)
Python source/version
Whether you used the installer app or manual install
Any logs/errors (screenshot is fine)
Thanks in advance—feedback will directly shape the next release.
Yes — a Chinese version is possible. I believe it should already support Chinese, but I haven’t tested it yet. I’ll run a proper test soon.
Would you be available for a short chat (in English) or by email so we can coordinate the localized glossary and a small user manual?
Hi everyone — I’m improving the default “bootstrap prompt” used by my MCP setup (the short instructions the LLM reads first, e.g. AGENTS.md). The goal is to make tool usage more reliable and reduce “LLM guessing”.
If you use this kind of workflow with Glyphs (or would like to), I’d love your input:
What I’m looking for
Use cases: 1–3 concrete things you’d want an AI + tools to do in Glyphs(ex: generate missing kerning pairs, spacing checks, stems/metrics adjustments, batch fixes, reporting, etc.)
Prompt suggestions: what should be explicitly stated so the LLM doesn’t do stupid stuff?
Failure modes you’ve seen: where do assistants usually misunderstand the task?
If you can, include
goal + steps
which data/files it should read
what the output should look like (report, diff, list of actions, etc.)
Thanks — I’ll use the best ideas to improve the default prompt and share the updated version back.
Now it is simpler to use it also in the default Claude app.
I’m currently adding a proper translation/localization layer in the UI. As soon as it’s ready, I’ll extract all the UI strings and email them to you so you can translate them into Chinese.
For the documentation, I’ll try to produce a first Chinese version with an LLM, and if you can review/correct it afterward, that would be amazing.
Thanks again in advance — really appreciate your help!
macOS plug-in. I’m still working on how to use it. No problem with installation.macOS plug-in. I’m still working on how to use it. No problem with installation.