The Frontguard
brand system.
A terminal-native identity for an open-source developer tool. Sharp, precise, mono-forward — built to look at home in a CLI and a PR comment alike.
01 / THE MARK
A geometric shield = protection, the guard at the gate. Built from a single 5-point polygon — no decorative SVG.
A center split divides the shield into two halves — baseline vs current. The visual diff, built into the mark.
The wordmark ends in a blinking block cursor — the terminal, always.
02 / COLOR
oklch(0.72 0.18 50)
03 / TYPOGRAPHY
04 / VOICE
Lead with the real problem — false positives and flake. Skeptical engineers smell hype. Don't oversell.
Specifics over adjectives. "Restore flex-direction: column at <768px" beats "fixes your layout."
The wordmark and commands stay lowercase, like the CLI. Terminal-native, never shouty.
05 / MESSAGING
Everyone adds visual regression tests. Then everyone mutes the channel they post to. Frontguard uses AI vision to tell a real regression from noise — so a red run means something again.
- Name the real problem: false positives, flake, muted channels.
- Quote the classifier: "intentional change, not a regression."
- Lead with BYOK, MIT, and self-hostable — earn trust with facts.
- Publish real numbers, including where it gets things wrong.
- Promise "zero false positives" or "100% accuracy."
- Call it magic, autonomous, or a silver bullet.
- Bury that AI is optional and runs on your own key.
- Shout in title case or pile on exclamation marks.