Project node

What it is. The root of one app. It holds every other node and is the first thing your agent reads when it opens your project.

When to use it. Create one per app you are building, then give it a name and a short line on what the app does.

What to use it for.

• Naming your app and describing its purpose in a sentence
• Linking the GitHub repo your agent will build into
• Connecting your AI agent so it can read the whole map

Feature node

What it is. One main part of your app, like a login page, a dashboard, or a checkout.

When to use it. Add one for each core thing a user can do, and keep the list short so the map stays focused on the problem you are solving.

What to use it for.

• Splitting your app into the handful of parts that matter
• Giving each part a name and a one line description
• Hanging sub features, notes and a PRD Bridge underneath it

Sub feature

What it is. A screen or flow that lives inside a feature, listed as a set of cards.

When to use it. Break a larger feature into the individual screens a user moves through, so the agent builds them one at a time.

What to use it for.

• Listing the screens in a feature, like sign up, sign in and reset password
• Breaking a long flow into steps the agent can build in order
• Keeping a big feature readable instead of one vague block

Bridge node

What it is. A small hub that branches off a feature so you can fan out and add more nodes to it in one place.

When to use it. Reach for it when a feature needs extra context, like a note, a mood board, or a PRD, and you want it grouped tidily.

What to use it for.

• Fanning a feature out into the extra nodes it needs
• Keeping related context under one connection point
• Adding a note or a mood board without cluttering the feature card

Notes

What it is. A plain language description attached to a feature, marked with a green tab.

When to use it. Use it to tell the agent how something should behave, in your own words. It is the cheapest way to add context.

What to use it for.

• Spelling out a rule, like passwords must be at least eight characters
• Describing how a screen should behave when something goes wrong
• Leaving context you would otherwise repeat in chat every time

PRD Bridge

What it is. The node that turns a mapped feature into a plan, marked with purple tabs. It writes your front end and back end PRDs together.

When to use it. Reach for it once a feature is mapped and you are ready to generate the plan your agent will follow.

What to use it for.

• Generating a feature's front end and back end PRDs in one step
• Keeping the two sides of a feature consistent from the start
• Choosing which kind of PRD to create: frontend, backend or database

Frontend PRD

What it is. The plan for what a user sees, covering the screens and how the interface should look and behave.

When to use it. Generate it from the PRD Bridge when you want the agent to build the visible half of a feature.

What to use it for.

• Describing the screens, layout and states of a feature
• Pointing the agent at the look and feel from your mood board
• Giving the front end a brief that matches the back end plan

Backend PRD

Project
  id          uuid  pk
  org_id      uuid  fk
  name        text
  created_at  timestamptz

What it is. The plan for the logic behind the screens, the data it reads and writes, and how the pieces connect.

When to use it. Generate it alongside the front end PRD when a feature needs real behaviour, not just a layout.

What to use it for.

• Describing the logic and rules a feature runs on
• Setting out the data it stores and the shape of that data
• Giving the back end a brief that matches the front end plan

Less common nodes

You will not reach for these on every feature, but they earn their place when a screen needs a look, a shape, or a layout pinned down. They work exactly like the core nodes, so they are tucked away here to keep the map above uncluttered.