← All work
Prompt Library — a Supabase-backed prompt library built in Figma Make
An internal prompt library for the design team — disciplines, tags, search, sign-in. Built in Figma Make with three prompts.
Internal team tool · 2026 · live

Three prompts, one team prompt library — written in the same framework it stores

Years Q1 2026 · live built alongside Memorisely AI Design + AI Coding
Stack Figma Make · Supabase (CRUD + auth)
My role Solo · three prompts to ship the whole app
In one line The library and the prompts that built it use the same RTCCF shape.
01 · The brief

Designers don't run out of ideas — they run out of structure. Without it, the first reply is surface-level and we burn tokens fixing it.

The team writes prompts every week, but each person writes them differently — same designer, different shape every time. So I picked two frameworks for the team to share: RTCCF (Role, Task, Context, Constraints, Format) as the front-loaded shape for the first prompt, and PRISM for the follow-ups that come after — review what's been built, iterate on one thing, specify exactly what you want, maintain what's already working. Then I built the library to hold them — in Figma Make, with three prompts, in the same RTCCF shape end to end.

The rule it's built on

“Front-load Role, Task, Context, Constraints, Format — and the first reply is already most of the way there. The library stores every prompt in that same shape.”

02 · What's inside

A working library — five disciplines, search, tags, sign-in

03 · Built with three prompts

Three prompts. One library. RTCCF end to end.

I wrote the three prompts in the same RTCCF + PRISM shape I wanted the team to use. The library then stores team prompts in that shape. The Figma Make file below has the full AI chat history — every wrong turn included.

Prompt 1 · UI + behaviour

The whole library, in one RTCCF prompt

One long prompt covering the role (senior product designer + front-end engineer in Figma Make), the task (every screen and interaction the library needed), the context (internal tool, RTCCF as storage shape, clean SaaS aesthetic), the constraints (field order fixed, search across title + discipline + tags, scope = CRUD only, mobile-responsive) and the format (a working Figma Make app).

Read the full prompt
ROLE
You are a senior product designer and front-end engineer building a polished, production-quality internal web app in Figma Make. You specialize in clean SaaS interfaces with strong information hierarchy and thoughtful interactive states, and you connect front-ends to a Supabase backend for data persistence.

TASK
Build a "Prompt Library" web app where a design team stores, browses, and copies reusable AI prompts organized by discipline, backed by a Supabase database for full CRUD (create, read, update, delete). I want to use supabase to create very basic CRUD database to create, read, update and delete prompts. Nothing more needs to be added just this simpe function. The app must include:

A left sidebar with the workspace name/logo at top and a navigable list of disciplines (UX Research, Writing, Design Systems, Prototyping, Design Engineering), each with a short subtitle and icon. Settings and Help & Support pinned to the bottom.
A top bar with a search field ("Search prompts, disciplines, or tags…"), a hint ("Click cards to view details"), and a primary "+ Add Prompt" button.
A main content area showing the selected discipline title, a prompt count, a horizontal "Filter by tags" row of pill chips, and a responsive grid of prompt cards.
Prompt cards displaying: title, tag pills (with a "+N" overflow chip), a ROLE section and TASK section preview (truncated), a "Click to view full prompt details" link, author name, date, and a "Copy Prompt" button.
A clickable card that opens the full prompt detail (all RTCCF sections) with Edit and Delete actions.
An "Add New Prompt" / "Edit Prompt" modal with fields: Prompt Title*, Author*, Discipline (dropdown), and an RTCCF Framework section with a short helper line ("Structure your prompt using the Role, Task, Context, Constraints, Format framework") followed by labeled textareas for ROLE*, TASK*, CONTEXT, CONSTRAINTS, FORMAT, plus a tags input. Save and cancel actions.

CONTEXT
This is an internal tool for a product design team that wants to standardize prompt-writing using the RTCCF framework (Role, Task, Context, Constraints, Format). Every prompt is authored in that structure and persisted in Supabase. The aesthetic is clean and minimal — a modern SaaS dashboard: neutral light background, generous whitespace, subtle card borders, rounded corners, a dark primary button, small uppercase section labels (ROLE, TASK) in muted gray.

CONSTRAINTS
Use the RTCCF field structure exactly: Role, Task, Context, Constraints, Format — in that order.
Search filters across title, discipline, and tags. Tag chips filter the visible cards. Discipline sidebar selection filters which prompts load.
"Copy Prompt" copies the full assembled prompt (all filled RTCCF sections, labeled) to the clipboard with a brief confirmation.
Single cohesive light theme, mobile-responsive (sidebar collapses, grid reflows to one column). Keep scope to CRUD only — do not add features beyond this.

FORMAT
A working, interactive web app in Figma Make connected to Supabase. Supabase should be used to create very basic CRUD database to create, read, update and delete prompts. Nothing more needs to be added just this simpe function.
Prompt 2 · Persistence

Wire Supabase for basic CRUD

The follow-up — exactly:

Supabase should be used to create very basic CRUD database to create, read, update and delete prompts. Nothing more needs to be added just this simpe function.
Prompt 3 · Team auth

Add team-only sign-in

The third and last — exactly:

Now we have added CRUD operations using supebase we want to go ahead and add some basic user auth to ensure it's just out team accessing and using our prompt system.
Supabase Table Editor — the key-value store holding every saved prompt
The Supabase project the second prompt set up — a key-value table holding every saved prompt. Each row is one stored RTCCF prompt.
The library in action — disciplines, search, prompt detail, Copy Prompt. Three prompts in, working app out. The Figma Make file with the full AI chat history is one click away in the credentials below.
04 · Try it live

Sign in and have a look

Copy the credentials below, open the URL, and you're in — the same library the design team uses every day. Browse the disciplines, open a prompt, see RTCCF in the wild. The Figma Make link at the bottom opens the build file with the full AI chat history if you want to see how it was made.

Prompt Library sign-in screen — Design Team · Internal Tool, email + password fields, Sign In CTA
URL
prompt-library.figma.site
Email
email@gmail.com
Password
fv1sP5mS20Q

Next: Research with AI

See the methodology that uses these prompts →