The problem
Teams jump to screens before they agree the problem.
Findings lose their link to what users actually said.
Accessibility gets checked last.
02
User-centred design, Discover to Deliver. Your yes at every stage.
UCD Pilot turns a brief or transcript into research themes, a problem statement, user stories, a journey, a UI plan and a working GOV.UK prototype. Every finding links back to what users said.
Or get all nine agents in the UCD Team bundle.
Teams jump to screens before they agree the problem.
Findings lose their link to what users actually said.
Accessibility gets checked last.
Builds an evidence log and themes, each tied to a quote
Writes a solution-free problem statement, assumption map, How Might We questions and a testable hypothesis
Writes user stories with Given-When-Then and accessibility criteria
Maps the journey and plans pages with GOV.UK Design System components
Builds a prototype and runs WCAG 2.2 AA checks before handover
Get this on Gumroad and download the zip.
Unzip it into a project folder and open it in your AI tool.
Say `run the UCD pipeline on <file>`.
It works one stage at a time and stops for your yes.
Evidence map: themes, each tied to what users said
Problem statement, assumption map, How Might We questions, hypothesis
User stories, screen content, journey map, UI plan, design tokens
Working GOV.UK prototype, accessibility check, handover notes
GDS Service Standard and WCAG 2.2 AA.
Every claim cites its source. Nothing is written to your tools without your approval.
Delivery managers and product owners without a UCD lead to hand, and prototypers who want a disciplined start.
An AI tool that reads project instructions: Claude Code, Cursor, or a tool that reads AGENTS.md. Claude Projects and custom bots work too, with copy-and-paste outputs.
Your own AI subscription or API key. Nothing to host.
Optional connections: Figma, Miro, Google Drive, web search, axe-core.
| File | What it does |
|---|---|
00-root-instructions.md | Who the agent is, the rules, the gates |
01-discover.md | Stage 1: Discover |
02-define.md | Stage 2: Define |
03-develop.md | Stage 3: Develop |
04-deliver.md | Stage 4: Deliver |
05-tools.md | Connections and permissions |
06-manifest.md | Which skill uses which tool |
07-adapters.md | How to load it in your tool |
LICENCE.md | What you may and may not do |
README.md | Start here: set-up in two minutes |
guardrails.md | Risks, controls and sign-off |
memory.md | What is kept and where |
adapters/AGENTS.md | Adapter for your AI tool |
adapters/CLAUDE.md | Adapter for your AI tool |
adapters/chat-system-prompt.md | Adapter for your AI tool |
adapters/ucd-pilot.mdc | Adapter for your AI tool |
evals/eval-suite.md | Tests and pass marks |
samples/01-passport-eligibility.md | Synthetic sample input |
samples/02-epc-rating.md | Synthetic sample input |
samples/03-universal-credit.md | Synthetic sample input |
samples/04-driving-licence-70.md | Synthetic sample input |
samples/05-council-tax-reduction.md | Synthetic sample input |
Designed before it was built
This agent was designed with the five-question method: purpose, boundaries, what it touches, when a person steps in, and how you know it works. [Read the design](./anatomy).
Designed by Vijay Kandhalu.