A linter for brand guidelines
A brand team at a mid-size agency runs every outgoing deck and asset through a bot that checks it against the client's brand guidelines — wrong logo lockup, off-palette colors, banned phrasing — before anything leaves the building. Violations come back as a checklist, not a bounce.
Guideline enforcement is the least glamorous, most reputation-saving job in branding — and it automates almost completely.
The how
- Each client’s guidelines PDF is distilled once into a structured rule file (lockups, clear-space, palette values, voice rules, banned terms).
- An n8n flow watches the “outgoing” folder; new PDFs/decks get page-imaged and sent to Claude with the client’s rule file.
- Claude returns violations with page numbers and severity; borderline calls are flagged “judgment” rather than auto-failed.
- Report lands in the project channel. Humans overrule freely — the bot is a checklist, not a gate.
Run it back
One afternoon per client to structure the rules, one n8n flow forever. Start with your most guideline-sensitive client and the most embarrassing past mistake.