Project dashboard
Settings
Everything you can turn on and off in a project lives in Settings, at the bottom of the project menu. There are four tabs: General, Automations, Status line and Design. Each switch is the same one you'll find on the screen it comes from, so changing it here changes it there too. Only people who can edit the project can change settings. Viewers can only look.
General
- Content language: everything Faundr records for this project (Overview, Stack, features, decisions, findings, handoff notes) is written in this language, whatever language each person chats in. Until someone picks one, it shows Not set yet, and it gets set during the next agent session from the language of whoever's working. When you change it, the dashboard checks with you first ("Write in Português from now on?"). Only new content follows the new language. The confirmation reminds you that rewriting existing content (Overview, Stack, Architecture, design guide, Showcase) uses tokens from your AI plan, and lists the command for each one along with what it used last time. Teammates who read the dashboard in another language can use Translate at the top of the page. See Overview and Stack.
- Economy mode: the quick design audit and the Overview and Stack updates stop running automatically at the end of responses and only run when you ask. The card shows what end-of-response checks cost over the last 30 days (their share of what this project's sessions spent, how many times they ran and roughly how many tokens each took). Proof checks, the handoff note and security checks always keep running. It's the same as running
faundr setup --economy allin the terminal, and the steps it turns off are labeled "Part of economy mode" under Automations. - Agent behavior:
- Reminder not to write too much code (the same switch as in Quality): at the start of each session, 6 lines telling the agent to reuse what already exists, skip premature abstractions, and never cut validation, important error handling, security or tests. It uses a little context. It can only be turned on after the quality check has run once (
/faundr:quality-scan). - Fit new folders into the architecture automatically (the same switch as in Architecture → Off-plan): when it's on, the agent puts a new folder in the most likely part and tells you where it went. When it's off, the agent suggests a part and waits for your OK.
- Connect your AI agent: the
/faundr:linkcommand with this project's ID, ready to copy. Run it once in your agent, opened in the code folder. After that, every session opened in that folder shows up live in Activity.
Automations
What Faundr does on its own at each moment (session start, every request, end of response and inside Faundr's skills), with one switch per step. When a step is off, it stops running for the whole team starting with the next request (session-start steps, from the next session). These are the same switches you'll find on each workflow's diagram under Workflows → From Faundr. A few always stay on: security checks and asking the owner before changing anything important.
Status line
The status line is the line Faundr shows at the bottom of your agent in the terminal. It tells you what's being built and, if you want, how much of the session context and your plan's limits you've used. It's only available in Claude Code, since Codex doesn't let plugins change its screen.
> _ Faundr · Shop · Free shipping 2/5 · now: Shipping tests · ctx ███░░ 42% · 5h ████░ 84% resets 18:40 · wk ████░ 61%
To turn it on for this computer: !faundr statusline-install. If you already use another status line, it's only replaced when you add --force.
What can appear
| Item | What it shows |
|---|---|
| Project | The name of the project connected to the folder. |
| Current feature | What's being built and how many steps are done (2/5). |
| Current step | The step in progress, or what's next. |
| Code map | Whether the map is current or updating. |
| Concerns | How many are open. |
Context (ctx) | How much of the context window is used. When it fills up, the agent compacts the conversation and loses detail. That's the time for /compact or a new session. |
5-hour window (5h) | How much of your 5-hour usage limit you've used and when it resets, on Pro and Max plans. |
Week (wk) | How much of your weekly limit you've used and when it resets, on Pro and Max plans. |
| Session cost | What the session would cost at API rates. Off by default, since Pro and Max plans aren't billed per token. |
| Bar above the prompt | The context and limit bars, also shown in a strip right above where you type. Off by default, and it needs the Faundr mod (see below). |
Each bar has 5 blocks, and each block is worth 20%. Context and limits only show up after the first response of the session.
Choosing what appears
Under Settings → Status line, each item has a switch, and the Preview shows how the line will look. Click Save to keep your changes, or Undo to go back. Your choice applies to everyone on the project, starting with each person's next message, with no restart needed. Anything that doesn't fit the terminal's width wraps to a second line.
Warnings
With Warn when usage is high on:
- at 80%, the item turns bold and gets a hint: "time to /compact" for context, and the reset time for limits;
- at 95%, the item is highlighted (inverted background): "use /compact now", or a note that you're about to hit the limit.
The status line updates itself after each of the agent's responses, without using tokens or the network.
The Faundr mod (optional)
faundr-mod is an add-on for Claude Code itself. It pops up a quick notice in the terminal when context or a limit passes 80% and 95% (once each), and it draws the Bar above the prompt when that item is on. To install it, run this in your agent: /plugin install faundr-mod@faundr. See Mods.
Design
The design scope (whether it needs to work on phones, the product type, the theme and motion), your approved exceptions and the disabled rules, each with a Turn back on button. It's the same as the Design section.
The terminal bar is the line Faundr shows below your AI agent: what is being built and, if you turn them on, the session context and your plan's usage limits. Choose what appears in the Status line tab. Run these commands in your AI agent, from the folder linked to the project.
The bar updates itself after every agent response, with no tokens or network calls: it reads what the agent already knows and the summary Faundr keeps in the folder.
/faundr:setup [--refresh]Sets up Faundr in the project in one go: checks Node, git and login, links the folder, builds the code map, the Stack and the first checks for free (no AI), turns on the bar and the skills, and shows what each AI-powered item costs so you choose. --refresh redoes the map and the checks.
e.g. /faundr:setup --refresh
/faundr:statusline [--force]Asks the agent to turn on the Faundr bar (the same as the command below).
!faundr config lang [en|pt|auto]Chooses the language of what the plugin shows you (setup, status bar, findings). Without a value, shows the current one; auto follows the computer's language.
e.g. !faundr config lang pt
!faundr statusline-install [--force]Turns on the Faundr bar on this computer. If another bar is already set, it only replaces it with --force.
/faundr:mod [what the mod does]Creates or adjusts an agent mod (side panel, band above the prompt, notice, command, rule that blocks a tool) following best practices: it asks the essentials, validates, type-checks and tests before handing it over.
e.g. /faundr:mod a notice when I go more than 30 minutes without running the tests
/plugin install faundr-mod@faundrInstalls the Faundr mod: alerts at 80% and 95% of context and usage limits, plus the band above the prompt (if turned on here).