Product design, AI pipeline spec, brand system, and a live consulting workspace - delivered in the first engagement phase. The client has a working product they can review, challenge, and ship from. Not a deck.
Node/Express tenant on FixBroken OS with scoped overrides. Multi-step guided capture to AI pipeline spec. Static workspace with timestamped deliverables. Hosted as a tenant subsite on the same Lightsail platform.
End-to-end product flow. AI-generated content from user input. Full brand identity and voice system. Live consulting workspace with audit trail and deliverable tracking.
Node/Express on Lightsail. FixBroken OS design system with scoped tenant overrides. Static workspace with deliverables timestamped to the second.
Not a pitch deck. The client sees the product flow, reviews deliverables, reads the audit - all live. Every deliverable has a ticking clock showing time since delivery.
FixBroken OS, the toolchain that enforces it, and the multi-tenant platform running underneath. The system this page is built on.
Single-file CSS design system with machine-readable JSON manifest. 15 Node.js CLI tools - voice linter, 6-category page audit, WCAG contrast checker, tenant scaffolder, manifest generator. 4 Express tenants on one Lightsail box behind nginx, systemd, webhook auto-deploy.
One CSS file serves every tenant. 14 token categories, 29 components, 11 organized sections. Overrides happen per-tenant via scoped wrapper classes - never by forking. Machine-readable manifest drives the toolchain.
Voice linter catches banned phrases and CSS violations. Page audit scores 6 categories. Tenant scaffolder generates nginx config, systemd unit, deploy script, and override CSS in one command. Pre-commit hook blocks violations.
AWS Lightsail, Ubuntu 24.04. Each tenant is an Express server on its own port, managed by systemd, proxied by nginx. Push to main, live in 10 seconds. Real-time token telemetry at /tokenz.
Free diagnostic tools built on the same engine we use internally. Paste a design system, brand doc, or copy - get a scored, shareable result in 30 seconds. Live at fixbroken.ai/free.
Browser-based tools backed by the FixBroken OS audit engine and voice linter. POST to /api/audit with raw HTML, get a JSON scorecard. Voice lint checks copy against the manifest ruleset. No auth, no signup - paste and run.
Paste HTML, get a 6-category scorecard - voice, design, structure, accessibility, responsive, performance. Letter grade, pass/warn/fail per check, actionable findings.
Paste copy or markup, get flagged for marketing sludge, banned phrases, emoji violations, and CSS namespace issues. Rules read from the machine-readable manifest.
The tools demonstrate the system. If the audit catches problems you care about, the conversation about fixing them is a natural next step.