Who it is for
Content leads who need every piece checked before it ships.
What it does
- Each run records its brand context, research, strategy and outline step by step.
- A person approves each stage before the next one starts.
- The editor scores how human the draft reads, and the finished piece is signed off before it ships.
How we built it
ArcLayer did the product design, the design system and the engineering. AI agents wrote code and tests against WriterDean’s design system, and a person reviewed each change before it merged.
Build note: Why every product gets its own design system
How we work on WriterDean
Every gate leaves a log
Build, type checks, tests, end-to-end tests, accessibility checks and evals each write a log to the repository.
Accessibility to WCAG AAA
axe-core and the Storybook accessibility addon check components and pages against WCAG AAA.
Writing quality is measured
Golden and humanizer evals score the output, and a lint pass checks the interface strings.
Writing skills are locked
Each writing skill is checked against a hash lock. An update rebuilds the lock and waits for a person to commit it.
Tests run without real accounts
End-to-end tests run on an embedded Postgres with the model calls mocked.
Secrets are scanned
A gitleaks scan checks the repository for secrets, and migrations run through one command against Supabase.