FaundrDocsOpen dashboardDashboard

Project dashboard

Overview and Stack

These two sections answer the most basic questions: what this project is (Overview) and where it runs and what it's built with (Stack). They're useful for you, for a co-founder who just joined and for a developer about to take over the code.

How a project's dashboard is organized

Before the sections themselves, here's how to get around inside a project:

  • The sidebar starts with Insights (the page that opens with the project) and sorts every other section into three groups, by what you're there to do:
  • Project, to understand the project: Overview, Architecture, Stack and Code map.
  • Work, to keep up with the work: Features, Memory, Workflows, Activity and Protection.
  • Quality, to check things before you launch: Design, Quality, Tests, Errors, Security and Launch.
  • Settings sits on its own at the bottom of the sidebar, below the groups.
  • Each section has tabs at the top. Tabs that hold collections show a count, like how many issues are open.
  • Every section ends with a Commands tab: the commands for that area, what they do and a ready-to-copy example.
  • The section and tab are part of the page URL, so you can send a teammate a direct link to a tab.
  • Almost everything updates in real time: you can watch the agent work without reloading the page.
  • When the project's content language differs from your dashboard language, a Translate button appears at the top of the page. See below.
  • Viewers see a notice at the top, and the buttons that change things are hidden.

Content language

Everything Faundr records for a project (Overview, Stack, architecture, features, decisions, findings, handoff notes, design.md) is written in a single language, the content language: English or Portuguese. It applies to the whole project, whatever language each person uses to chat with the agent. The agent still replies to each person in their own language.

  • Where to set it: Settings → General, on the Content language card. A new project starts with the language of your dashboard. An older project with no language set gets one on the next agent session, based on the language of whoever's working.
  • Changing it later only affects new content. What already exists stays as it is. To rewrite something in the new language, ask the agent (/faundr:overview, /faundr:stack, "draw the project architecture", /faundr:design, /faundr:design-showcase). Each one reads the project again and uses tokens from your AI plan, about as much as the first time it ran. The card shows how much each one used last time.
  • Translate: anyone reading the dashboard in the other language can use Translate to English (or Translate to Portuguese) at the top of the project page. Translation runs in your browser's built-in translator (Chrome), on your computer, without using tokens. Code, file paths and commands stay as they are. The first time, your browser downloads the translator and the button shows the progress. To go back to the original text, click Show original. The button only appears when your browser has the translator.

Overview

The Overview is a summary of the project for people who don't code. The agent writes it from the code and docs themselves when you run /faundr:overview.

Tabs

TabWhat it shows
SummaryWhat the project is, who it's for and what problem it solves. It also shows how many parts are done and in progress, along with the questions still open.
How it worksThe numbered steps of how it works, with the files that back up each one.
System partsA card per part (screens, API, server, data, integrations…), each tagged Done, In progress or Planned.
Current statusThree columns: Next steps, In progress and Done. Next steps and In progress come live from the Features board. Done is what the agent wrote in the Overview.
DecisionsThe project's key decisions and the reason for each.
How to useHow to run and use the product, with commands where there are any.
GlossaryThe project's terms and what they mean.

The Summary also shows when the Overview was written and at which commit.

Architecture

Architecture has its own item in the project sidebar, Architecture: an always-open canvas you can zoom and drag like the dashboard's other canvases. It shows the parts of the system and how they talk to each other, in layers: users, screens, server, data and outside services. It's optional and appears when you ask the agent for it, at the start of the project or with "draw the project architecture".

  • Click a piece to open its explanation: what it is, what it receives from and sends to, where it lives in the code and whether it exists yet.
  • Planned vs. built: at the start of every session, the plugin checks on its own (without AI) whether each part already exists in the code. Parts that are still only a plan appear dimmed, and the header says how many parts exist. The canvas updates itself.
  • Off-plan: the button in the corner lists folders with code that don't belong to any part (excluding tests, docs, configuration and images). When a new folder appears, the agent mentions it in the terminal and suggests which part it belongs to; the architecture only changes with your OK. If a folder is off-plan on purpose (one-off scripts, for example), click Intentionally off-plan on it, with a reason if you like. It leaves the list right away, the check stops flagging it, and it moves to a collapsed list showing who marked it, the reason and an Undo button. This only affects Off-plan: the folder still appears in the Code map and in the checks. To exclude a folder from everything (third-party code, generated files), use .faundrignore. Only people who can edit the project can mark or undo.
  • Auto-place new folders: a switch in the Off-plan panel, off by default. When it's on and a new folder appears, the agent puts it in the most likely part of the architecture without asking and tells you in one line where it went (if no part fits, it suggests one instead of creating a new part). When it's off, the agent only suggests and waits for your OK.

Starting a new project

If there's no code yet, describe your idea to the agent or use /faundr:planning <idea>. It asks a few product questions (who uses it, the key moment, what goes into the first version), shows you a summary to approve and, once you approve, writes the Overview from the conversation. System parts go in as Planned, milestones go to Features (the first one as current) and, if you want, the architecture gets drawn under Architecture in the project sidebar. As code appears, /faundr:overview updates the Overview from the real project.

Keeping the Overview up to date

You don't have to remember to update it. When the project changes enough for the Overview to fall behind, Faundr asks the agent to update it at the end of a response (once per session). To force an update at any time, run /faundr:overview.

With economy mode on (under Settings → General), the Overview and the Stack stop updating on their own at the end of responses and only update when you ask. See Settings.

A project without an Overview shows "This project doesn't have an Overview yet.", with the command to create one.

Overview commands

Your AI agent writes the Overview from the project itself. Run these commands in your AI agent, from the folder linked to the project.

At the end of a response, if the Overview is out of date, Faundr asks the agent to update it.

  • /faundr:overview

    Writes or updates the project Overview: what it is, who it is for, how it works, its parts, current status, decisions and how to use it.

Stack

Stack shows where each part of the project runs and what it runs on. It has two layers:

  • The automatic read, without AI, done at the start of every session: packages, deploy configuration files and environment variables (names only).
  • The AI explanation, in plain language, when you run /faundr:stack. It fills in what the files don't say. Tell the agent, for example, where your domain is registered or that deploys happen automatically from GitHub.

Every tab has a status line at the top: "Reviewed by your AI agent on…" or "Read automatically from your files, without the AI explanation". If the files changed after the review, a Changed since review tag appears, saying what was added and removed, with a button to update.

Tabs

TabWhat it shows
Where it runsA card per layer: screens, server, database, login, files, domain, deploy (how code goes live) and scheduled jobs. Each card shows the provider, what runs there, the address and the file that proves it. Below them are the Environments (production, staging, local), with address and branch.
Languages and toolsThe project's languages by share, the runtime version and the libraries, grouped by role (core, UI, styling, data, testing…).
What Faundr coversThe project's languages and what Faundr does in each one (map, quality, security, tests, design…). The full list of languages is in Supported languages.
External servicesThe outside services the project uses (payments, email, AI, monitoring, maps…), what each is for and which variables it uses.
VariablesEnvironment variables, names only. See below.
How to runThe project's commands (npm run dev, npm run build…), each with what it does.

Variables: is each one where it needs to be?

The most common reason an app "works on my computer but not in production" is an environment variable missing somewhere. The Is every variable where it needs to be? card checks four places for each variable:

ColumnQuestion
Read by codeDoes the code use this variable?
In exampleIs it in .env.example (or equivalent), so newcomers know it exists?
On your computerIs it in your local .env?
In productionIs it set on your hosting (Cloudflare, Vercel, Netlify)?

It flags these problems:

  • Breaks in production: the code uses it, but your hosting doesn't have it. The published app will fail.
  • Missing on your computer: the code uses it, but your .env doesn't have it.
  • Missing from example file: it exists, but nobody documented it.
  • Unused: it's set, but the code doesn't use it anymore.

The In production column only appears after !faundr env-check --hosting, which queries your hosting through the provider's CLI (the CLI must be signed in on your computer).

In the variables list, the Browser tag marks variables that ship to the browser, which means anyone can see them (prefixes like VITE_ and NEXT_PUBLIC_). They should never hold secrets.

Only variable names leave your computer. Values (keys and passwords) are never read for upload.

Stack commands

The Stack is read from the project files automatically (without AI) and explained by your AI agent in plain language. Run these commands in your AI agent, from the folder linked to the project.

At the start of every session, Faundr rereads packages, deploy config and environment variables. If something important changes after the agent reviewed it, the agent is asked to update it at the end of the response.

  • /faundr:stack

    Writes or updates the Stack: where each part runs, languages, tools, external services, environment variables (names only) and how to run it. Tell the agent what the files don't say (e.g. where the domain is registered).

  • !faundr env-check [--hosting]

    Checks that every variable the code reads is in the example file, in your local .env and on your hosting (names only). With --hosting, lists the secret names on Cloudflare, Vercel or Netlify through the provider's CLI (you must be logged in). Runs automatically at the start of every session, without --hosting.

    e.g. !faundr env-check --hosting

  • !faundr stack-scan

    Rereads the files now, without AI, and shows what was detected (the dashboard updates right away).