The Claude Code Starter Pack: Become an Orchestrator in One Week
A batched series of 4 guides plus 3 templates that take you from 'AI curious' to 'shipping with agents' in about a week. The fastest path from designer-as-executor to designer-as-orchestrator, built around one working review skill.
Why this series exists
Every designer is hearing the same thing right now: you need to use AI, you need to use Claude Code, you need to be an “orchestrator.” Almost nobody shows you what that actually looks like on Monday morning.
This series does. In one week, reading about an hour a day, you will:
- Understand the shift from designer-as-executor to designer-as-orchestrator
- Build your first working review agent
- Learn how to structure projects so agents work consistently
- Dispatch multiple agents in parallel for a full multi-lens review
By the end, you will have 5 skills running in your home directory, a CLAUDE.md in your repo, and a review workflow that cuts review time in half.
The shortest path from “AI curious” to “shipping with agents” is one working review skill. Build that, and everything downstream gets easier. This series centers around that one loop.
The 4 guides, in order
The series is designed to be read top to bottom over about a week. You can also jump directly to the hands-on guide (#2) if you want to build first and read the theory later.
The 3 templates, paste-ready
The templates are the deliverables. Install them while you read the guides, and by the end of the week you have the full stack running.
The one-week plan
If you want a structured path, here it is. About an hour a day.
Day 1 (Monday) — frame the work
- Read From Executor to Orchestrator. 12 min.
- Pick the one recurring task you want to automate first. (For most, it’s UX reviews.)
Day 2 (Tuesday) — build your first agent
- Read Your First UX Review Agent. 30 min.
- Install UX Review Agent Starter.
- Run the skill on a real screen. Refine based on the output.
Day 3 (Wednesday) — use it on real work
- Use
/ux-reviewon 2 or 3 real screens from your current project. - Note which rules were too vague. Tighten them in the SKILL.md.
Day 4 (Thursday) — set up project context
- Read Context Is the Product. 18 min.
- Install Project Context Checklist in your current project.
- Fill in the brackets. Commit.
Day 5 (Friday) — go parallel
- Read Parallel Review System. 25 min.
- Install Multi-lens Review Pack. Paste all four skills.
- Dispatch the 4-lane review on one important screen. Review the combined report.
Weekend — sit with it
- Don’t add more tools. Sit with what you have built. Next week you will feel the leverage.
Reading paths by use case
Not everyone starts the same way. Pick the path that matches your situation.
”I have 30 minutes and I want to see the point.”
Read From Executor to Orchestrator (12 min) and skim Your First UX Review Agent. Install the starter template. Run it once. Decide from there.
”I already use AI tools but my workflow is ad-hoc.”
Skip to Context Is the Product and Parallel Review System. These will upgrade how you orchestrate.
”I lead a team and want to roll this out.”
Start with From Executor to Orchestrator to build a shared frame. Then adopt the Project Context Checklist as a team standard. Then introduce the review skills gradually.
”I am a consultant or freelancer and I want leverage.”
Do the full series in order. The compounding benefit per client is the reason this pack exists.
How this series fits with the Taste Series
The Taste Series teaches you to encode your design taste into skills so AI output stops looking generic.
This series teaches you to orchestrate agents so you ship more work with less friction.
They pair. Most readers should do the Starter Pack first because it is shorter, more concrete, and produces immediate value. The Taste Series is the deeper investment in quality that follows.
What you walk away with
By Friday of week one, if you followed the plan:
- 1 CLAUDE.md in your project repo (checked in, versioned)
- 5 SKILL.md files in your home directory
/ux-review(the all-in-one reviewer)/review-a11y/review-first-user/review-copy/review-edges
- 1 repeatable multi-lens review workflow that runs in under 3 minutes
- 1 concrete shift from executor-thinking to orchestrator-thinking
None of the skills go away. Every one of them compounds. You pay the setup cost once. Every screen you review next year costs 30 seconds of invocation and 5 minutes of reading a deduplicated, ranked report.
That is the whole bet. One week of setup, a year of compounding leverage.
After the Starter Pack
Once this series is running, three obvious next steps:
- Do the Taste Series to make sure the output your agents produce matches your design standards, not the internet’s average.
- Build project-specific skills. A
/copy-audittuned to your brand, a/component-doctuned to your design system, a/handofftuned to how your engineers work. - Watch for series 3 (the Consulting Starter Pack, for people running design consultancies). Coming next.
Start here
Try with Prompts
Ready-to-use prompts related to this guide
Series
1 item in this collection