We'd rather go deep
than go viral
An indie publication for technical builders shipping with AI agents. We publish what we run ourselves.
-
Ship only what we use ourselves.
-
Avoid hype; back fundamentals.
-
Keep learning, forever.
-
Respect feedback; think for yourself.
// the tools
The tools, shown working.
-
Watch a PRD turn into shipped code, chapter by chapter. MVP-builder
AI demos stop at the prompt.
Builds MVPs with an agent that verifies its own work before it commits.
Claude Code (plugin) · MIT
-
Watch four commands turn references into production-grade UI. Design-builder
Generic AI workflows produce generic UI.
Produces production-grade UI without the AI slop — real references on every screen.
Claude Code (plugin) · MIT
-
Watch an agent pull real palettes and patterns over MCP. Designlib-MCP
AI agents design from training data, not from references.
Feeds them real palettes, fonts, and patterns over MCP, on demand.
Model Context Protocol (MCP) · MIT
-
Watch one Claude Code session run as a coordinated dev team. Dev-team-agents
The context that wrote the code can't review it.
Runs one session as a coordinated team — specialists, enforced handoffs, independent review.
Claude Code (plugin) · MIT
-
Watch the agent write current iOS, not last year's APIs. iOS-toolkit
Claude Code's iOS knowledge is frozen at its training cutoff.
Keeps the agent on current iOS — references + Swift guardrails, so it stops writing last year's APIs.
Claude Code (plugin) · MIT
// the blog
The editorial line.
The methodology behind the tools — what we ship, why, and what we learned shipping it.
Read the blog// subscribe
Get the methodology behind the tools.
Walkthroughs, tutorials, and lessons we extract from real work. Free, no spam, unsubscribe anytime.
// subscribed
Check your inbox for the confirmation link.