UX · CX · BX
Claude resources
Get more out of Claude.
A working repo to enhance Claude, save tokens, and learn to use it well. Every command here is checked against the official docs.
A skill is a folder with a SKILL.md Claude loads on demand. Browse a directory, then verify each repo before installing.
MCP servers and connectors link Claude to Drive, Gmail, Calendar, GitHub and Slack. See code.claude.com/docs and docs.claude.com.
Drive Figma Desktop through its built in DevTools with a community CLI, no API key. Read: Into Design Systems. Tool: silships/figma-cli. It is third party and work in progress, so vet it before running.
A library of DESIGN.md specs you can hand Claude: styles.refero.design. Or write your own so the output builds to your system, not a generic one.
A CLAUDE.md in your repo gives Claude your patterns and rules, so first pass output is on brand instead of generic.
Use /compact to compress a long session, /context to see what is filling the window, and /clear to reset.
The Caveman skill trims tokens by roughly 75 percent.
Haiku for simple and fast, Sonnet for the everyday default, Opus for hard reasoning, Fable 5 for the ceiling. Using the smallest model that works is the biggest saving.
Each revision optimises from the model's own previous output and converges on the generic, so you burn tokens getting further from distinctive, not closer. Start fresh for original work. See the read on AI design convergence.
/cost shows the current session cost. /usage checks usage against plan limits.
/batch runs changes across worktrees and subagents (/agents) split work, which avoids re running the same context.
Start with docs.claude.com for Claude and code.claude.com/docs for Claude Code.
Clear instructions, positive and negative examples, step by step reasoning, and XML tags. See the prompting guide.
Anthropic runs free courses on prompting, Projects and certification. Find them via anthropic.com.
Quick start (basics, prompting, Projects). Head start (design, first skill). Go deeper (Cowork, teams, building with Claude Code). One topic per sitting.
Set up a Project properly (clear name, Role and Output instructions, the right files, the right model). Build a one time workspace with an about-me.md and an anti-ai-writing-style.md so you explain yourself once.
- ultracode is not a distinct effort level. Docs confirm it reports as
xhigh. Real levels: low, medium, high, xhigh, max. - Model references use the current line: Opus 4.8, Sonnet 4.6, Haiku 4.5, and Fable 5.
Cross checked against Anthropic's official Claude Code docs and several independent June 2026 references. Over 60 built in commands plus bundled skills are confirmed. A red dot means the item is plausible but was not independently confirmed, so availability varies by version, plan and platform. Type / in your own build to confirm.