Everything this site is made of

Five-agent pipeline that writes SEO metadata for every asset.

The most-used custom workflow on the site: five sub-agents (auditor, blog, content, linker, metadata) that audit, draft, and cross-link content, then write it to Supabase. Token usage and cost are tracked per run.

Corpus checks, snapshots, edit digests and SME queues for an exam bank.

A deterministic content pipeline for the CRSS recovery-support question bank: gated corpus checks, bank snapshots, edit digests and an SME review queue, with every build recorded as a pipeline run.

The bridge that pushes committed content into the live database.

Blog, research and projects content is Supabase-driven. The seeders upsert committed seed sources into the main project so content reaches the live site, then rebuild the semantic search index. Auto-runs on merge.

Autonomous mobile QA for the WebGL/Canvas2D game library.

Two rungs of game QA: a nightly LLM smoke pass that screenshots every game at a mobile viewport with software WebGL, and a deterministic per-PR Playwright gate that blocks merges on layout, obstruction or render regressions.

Versioned, authored workflow graphs executed node-by-node.

A general workflow engine: versioned graph definitions, run history, and per-node execution records. Triggers fire off a domain-event bus; operator nodes dispatch agent jobs.

A task queue for autonomous agents, with schedules and artifacts.

The queue behind the autonomous agents: queued/running/done jobs with verdicts and results, recurring schedules, and artifacts (screenshots, files). The site-health canary and playtest smoke both run on it.

One catalog that powers the marketplace and the public tools page.

A content-as-code registry of installable add-ons (analytics, social, productivity boards). The same source renders both the /profile/settings marketplace and the public /tools landing, so the two can never drift.

Sixty-plus client-side browser tools, each its own SEO page.

A registry of free, client-side browser utilities (converters, encoders, generators, formatters). Each tool is statically generated with its own metadata, FAQ and WebApplication JSON-LD. Nothing leaves your machine.

One declarative list of every destination and action on the site.

A single source of truth for the site’s stable destinations and actions, role-gated and grouped. The nav, footer sitemap, omnibar and concierge all read it — so they never re-derive their own lists.

A living, interactive map of how the whole system communicates.

A content-as-code system map: layers, surfaces-by-zone, integrations, and an interactive ecosystem graph of how they all talk. Public and indexed — the front door to understanding the platform.

One helper that gives every page consistent SEO and social cards.

A centralized metadata builder plus dynamic OpenGraph image generation. Every page calls one helper for canonical URLs, OG tags, Twitter cards and Bauhaus-styled social images — consistent SEO with no boilerplate.

The site-wide design language: CSS variables + JS tokens, dark-aware.

The visual identity in code: CSS custom properties plus a JS token object, with a light/dark theme that switches on a data attribute. Every non-Tailwind surface styles itself from these tokens.

Reusable JSON-LD schema builders for rich search results.

A set of JSX components — Person, Website, Article, Software, Research, Breadcrumb — that each emit a JSON-LD script tag. Pages compose them to earn rich results and breadcrumbs in search.

One-click shared boards — todos, weather, trips, countdowns.

The free multi-tenant boards anyone can spin up at their own link: shared to-do boards, family weather dashboards, trip planners and countdowns. Publishable to a public, indexed read-only view.