Library

What MCP is, how it differs from an API, and 8 steps to add an MCP server to the app you already have, shown on the one I built for findnewpath.

Checks your prompts, skills, memory and agent harness against Anthropic's Claude Fable 5.2 guide, and fixes only what you saw go wrong.

Build the design system in Figma before you code, and every screen is put together from it. What it is, why it pays off, and ten steps with Claude, from variables to audits.

When a feature needs a new screen, Claude designs it in Figma first and I review the frames before any code. Shown on a real Coral feature.

Your main agent plans and reviews; cheaper subagents write the code from precise specs, only where the work splits cleanly.

Let Claude Code open what it just built in your own Chrome, check it at desktop and phone width, and report what's broken.

Runs a full security audit of a codebase with separate agents that find and then try to disprove each bug.

A design review of your real screens, web or iOS: every note sourced, marked on a screenshot, with the fix.

Lets your agent search for and install skills when you ask for something it can't do yet.

An AI caterer talks over real screen recordings: the script, the avatar, the edit, and what broke.

Connect Claude Code to Figma, let it design the screen, fix it there, then build from the frame.

Anthropic's skill for writing, testing and improving your own agent skills.

An AI avatar talks over real screen recordings of your product: script, avatar prompts, clip check, edit, posting notes.

UI UX Pro Max: a searchable design database your agent checks before it builds a screen.

Pushes Claude toward distinctive UI instead of templated defaults.

Wires up a what's-new sheet and store-review prompt in a React Native app.

Postgres rules for schema, RLS, indexes and migrations.

Design vocabulary and audits for polishing an interface.