Everything we ship, in reverse chronological order. Verticals, blog posts, shared packages, fleet infra. Subscribe via RSS if you want push updates.
Shipped batches 28-30. Five more landings live per 15-minute release cadence, all using the shared Dangercorn template — same design system, same shared packages (dcst, dcch), same deterministic port assignment. Each is a real Flask app on GitHub, each has a landing page, each is eligible for a hosted Pro tier when demand surfaces.
Full SEO quality pass. Added SoftwareApplication JSON-LD with Offer arrays to every vertical landing, plus BreadcrumbList schema tracing Home → Products → {slug}. Completed Twitter Card tags, added og:site_name, theme-color, and Twitter @dangercorn attribution. Sitemap now covers /services, /homeschool, /one-pager, /changelog, and every live vertical. Fixed the roomcraft page's bare-bones head.
This page. A public log of what we ship, when we ship it. Paired with an RSS 2.0 feed at /changelog.xml for anyone who wants push updates. Part of the transparency stance — ship in public, show the work, let search engines see the recurring content signal.
Product spotlight on childmilestone. Private-by-default milestone tracker for families who don't want their kid's photos feeding someone's ad algorithm. Self-host, SQLite, export everything.
Long-form breakdown of bookcircle — SwipeNight-style voting for what the club reads next, AI-generated discussion questions, per-member reading history.
Why we built advisorlink: turn a year of homeschool work samples into an advisor-grade PDF in minutes instead of a weekend. Written for the families we know.
honeybees is a hive manager for hobby beekeepers. Log inspections, track queens, and upload a frame photo to get brood pattern + mite signal analysis via Claude Vision.
Why we built cheesemaking. Batch logs, aging vault temp/humidity tracking, tasting notes, and a Claude Vision assistant that reads the rind of your 21-day Brie. The first full product spotlight in the series.
Post on the Python f-string backslash gotcha that our CI pipeline caught before it hit production. A short war story about why we wrote the linting layer we wrote.
Why every Dangercorn vertical gets a deterministic port based on app name hash. It breaks at N=~10,000. We are at N=135. Simplicity wins for a while, and "a while" is often "long enough."
Shared Anthropic API wrapper used by every vertical that does AI. Retry with backoff, automatic model downshift on 429, JSON mode with code-fence stripping, cost logging by slug. One pip install, one import, one place to fix rate-limit bugs.
The process behind generating 130+ product landings without it looking like AI slop. Spec files, shared template, human-written JSON, machine-rendered HTML, human-reviewed output.
The Flask shell every Dangercorn vertical is built on. Single-password auth, Stripe billing wiring, SQLite helpers, Jinja base templates, deterministic port assignment. One pip install dcst and a new vertical can boot in 30 seconds.
The scaffolding sprint post. 31 verticals stood up, landing pages generated, GitHub repos initialized, port bindings recorded. Not 31 products — 31 starts. Most won't get demand. The ones that do get promoted.
Position piece on why we're building 220+ verticals instead of raising a Series A. Slow growth, sustainable revenue, no board, no exit clock. Named the philosophy out loud.
Why the home cheesemaker / book-club organizer / 3D-print-farm operator is a real market — just not a VC-scale one. Small audiences, devoted users, sticky products.
Every Dangercorn product is free to self-host. Why that's the pitch and not the concession. Source on GitHub, run it anywhere, your data stays yours.
The number isn't random. It's the count of niches we can template with a single shared stack. 220 concrete verticals, one build system, one design language, one deploy pipeline.
Kickoff of the fleet series. How seven specialized AI agents run on distributed hardware across a Seattle home network. Why "home lab" became "production fleet." Architecture decisions and what we would do differently.
The design system this site runs on. Gold on deep brown, DM Serif Display + DM Sans, the grid background, the hover micro-animations. Every vertical landing inherits it.
Cold email performance from the SiteLens rollout. 349 unique recipients, bounce rate analysis, reply intelligence, what "works" vs what gets filtered. Real numbers, no PR polish.
How Njord — our weather trading agent — finds mispricings in Kalshi markets using an ensemble forecast. Live scan runs every 5 minutes, bias-corrected models across 15 cities.