Safety model

Glyphs MCP is designed to keep you in control.

Golden rules

• Read before write: inspect first, then propose changes.
• Dry run before apply: preview what would change.
• Never auto-save: save only when you decide (save_font).
• Kerning groups first: prefer class/group kerning; add exceptions only when you have proof.

Useful guides:

• Kerning workflow
• Kerning tools
• Spacing tools

Confirm-gated mutation pattern (copy/paste prompt)

Use this template with your AI client to keep changes safe and explicit:

You are my Glyphs MCP assistant.
Rules:
- Never auto-save.
- Never mutate without a dry run first.
- Show me a short summary of the proposed changes (counts + biggest deltas).
- Wait for me to reply exactly “apply” before calling any tool with confirm=true.

Task: <describe your goal>

1) Inspect current state with read-only tools.
2) Run the appropriate apply tool in dry_run=true mode.
3) Stop and wait for approval.
4) If I reply “apply”, run the same call again with confirm=true.
5) If I reply “save”, call save_font.

Optional: lock down access

By default, Glyphs MCP is a local server (127.0.0.1) and does not require authentication. If you want extra protection (for example in a shared environment), the server supports:

Static auth token

Set GLYPHS_MCP_AUTH_TOKEN to require a token on every request.

The token can be provided either as:

• Authorization: Bearer <token>, or
• mcp-auth-token: <token>

Allowed origins (hostname allowlist)

Some clients/proxies include an Origin header. The server can reject unexpected origins via:

• GLYPHS_MCP_ALLOWED_ORIGINS

This is a hostname allowlist (in addition to the defaults 127.0.0.1 and localhost).

caution

Keep the server bound to localhost. Don’t expose it on a public interface unless you fully understand the security implications.

See also: Settings.