Most stores learn about a broken payment form from the sales that did not come. The numbers sit in five dashboards nobody opens on a phone, a slow day gets blamed on the ad, and a test gets called after twenty visitors.
One command starts the app with a sample store, so you see every screen before you connect anything.
One script line on your site reports visits, taps on buy and the payment steps. Sales arrive from Stripe or from one web address your shop calls, ad numbers from Meta, and if your store already keeps its own tables, it reads those.
Add it to your home screen, enter your PIN and turn alerts on. Ask your AI how the week is going, where buyers stop, which version wins, or to add a card, an alert or a tab.
A live view of who is on your site right now, room by room: the ad page, the shop, the tools, the library and the payment.
Two skins for that live view: a control room board in a dark terminal look for people who code, or an animated pixel-art store, switched with one tap.
A live feed: a sale and where its buyer came from, a tap on buy, a payment opened and left, a payment form that did not load, a new lead.
A dashboard in ten groups, from your monthly profit goal and what needs you now to sales, ads, visitors, emails, customers and costs.
Real sales per day next to ad spend and what the ad platform claims, by week, campaign, weekday and hour.
Where each visitor leaves the payment, step by step, and every person who opened it, one by one.
Which version of a test sells more, with an honest "too few to decide" when there is not enough to tell.
Card against PayPal, inside the Instagram and Facebook apps against a normal browser, and one buyer end to end.
Your AI builds what you look at: number cards any script can fill, lines on the feed, alert rules and your own tabs in the phone app.
Phone alerts for a sale, a new lead, a visitor who tapped buy and left, a payment problem, and any rule you set.
Every customer in one place: what they bought, how they found you and everything they did on your site, with your own notes and tags next to it.
Set a follow-up in one tap or by asking your AI, and see each day who is due.
Save the groups you care about, like everyone who opened the payment and left this week, and download them as a CSV.
Robots and your own visits stay out of the counts, and only money really charged counts as a sale.
It runs on your own computer or server. It works on its own data, or reads the events, purchases and mail tables your store already writes.
A payment robot that buys from your store in test mode and checks the purchase was recorded.
It runs where you install it, on a computer or server that stays on.
To count visits from your real site and to send phone alerts, the app needs a public https address: your own domain or a free tunnel. The guide shows both.
Sales are read from Stripe, or reported by your shop to one web address. Any other payment company needs that one call.
Ad numbers are for Meta ads.
The readings of list mails and of the support inbox need your own database connected. The guide says what each reading needs.
It shows, measures and alerts. It never changes your store, your ads or your prices.
Paste these into your AI after you unzip Control Room DNA (vibedna.ai/store/control-dna). The full install handbook is at vibedna.ai/installs/control-dna.md. It was called Control DNA until 2.0.0; a copy you already set up keeps its folder, keys, entry and addresses, so these prompts work for it too.
There are two prompts. Paste the first one once, right after the unzip: it sets up the app, shows it to you with the sample business, then connects your site, your sales, your ads and, if your store already writes its own tables, your database. Paste the second one at the start of any day you want your AI to read how the business is doing, or to build something on the app.
Prompt 1: setup (once)
I just unzipped Control Room DNA 2.2.1. It is a control app for my online business plus the readings that measure it. The app runs on my own computer or server and opens on my phone and PC: a live view of who is on my site right now (a control room board in a terminal skin, or an animated pixel store), a live feed, a dashboard grouped as Vision, To look at, Sales, Ads, Visitors, Emails, Customers, Costs, Achievements and Health, a People tab with one record per customer (what they bought and did, my notes, tags, follow-ups and segments), custom pages and phone alerts. Its MCP server gives you 42 tools to read my numbers the same way every day, to keep my notes and follow-ups on each customer, and to build my dashboard, number cards, feed lines, alert rules, app pages and its look. Set it up with me now, one step at a time, and show me each change before you make it. WHAT IT IS FOR Seeing the business live and measuring it the same way every day, so a change is judged by numbers and not by feel. Money counts only from my own sales records; the ad platform's own count is shown next to them, never in their place. STEPS IN ORDER 1. Read INSTALL.md in full before anything else. 2. Check Python 3.10 or newer. Run pip install -r requirements.txt. Ask me whether I want phone alerts; if yes, pip install pywebpush. 3. Run python -m controldna setup with me. I type the PIN myself; never ask me for it in the chat and never write it anywhere. Ask me for my business name, my currency and my monthly profit goal, and pass them as --name, --currency and --goal. The licence key goes in the keys file as MCPIZE_LICENSE_KEY, typed by me. 4. Run python -m controldna demo, then python -m controldna run, and tell me the address to open. Let me look at the sample business: the Live tab, the Dashboard tab and its groups, and the People tab with its sample customers (open one person). Show me the two buttons at the top right that switch between the terminal skin and the pixel skin on my device, ask me which one I want on every device, and set it with set_look(skin=...). When I say so, run python -m controldna demo --remove. 5. Add the "control-dna" entry of INSTALL.md section 11 to my AI app, with the full path to server.py. After a restart, run setup_status. 6. A public https address: explain the two ways of INSTALL.md section 10 (my own domain with a reverse proxy, or a free tunnel) in plain words and let me choose. Help me set it up, then save it with set_business(public_url=...). 7. My site: set_business(origins=[my site address]). Run tracker_snippet and help me paste the line on every page, mark my buy buttons with data-buy and my lead forms with data-lead, add the payment steps where my payment happens (pay_open, pay_ready, pay_touch, pay_fail, pay_close, the PayPal steps, purchase_return), and controlDNA.ab(...) if I run an A/B test. Have me open my site once with ?cdna_owner=1 so my own visits never count. 8. The rooms: look at my site's pages with me and set set_business(rooms=...) so each page lands in the right room of the store. 9. Sales: Stripe (a restricted key with Read on Charges, Checkout Sessions and PaymentIntents, which I write into the keys file myself as STRIPE_KEY) or my shop calling the sale hook. One way per shop. Ask me for my own email addresses and set set_business(own_emails=[...]) so my test purchases never count. 10. Ads, only if I run them: META_TOKEN and META_AD_ACCOUNT in the keys file, written by me. Ask me which countries my ads run in and set set_business(ad_countries=[...]). 11. Data source of the readings: ask me whether my store already writes events, purchases and mail_log tables in a database of its own (INSTALL.md section 8). If not, keep the local source and say which readings need more (setup_status lists them). If yes, I write SUPABASE_URL and SUPABASE_KEY in the keys file myself; then set the rest with set_data_source (thankyou_prefix, test_email_patterns, buy_steps, identity_raw_keys, access_lookup, access_labels, inbox as my store needs), and run setup_status until every reading I use says "works". If I used Measure DNA, its settings are read as they are: show me where each one came from. 12. Phone alerts: on my phone, open the https address, add it to the home screen, tap Turn on alerts here, then run send_test_alert. 13. Ask me which alert rules I want (a sale above an amount, no sale for some hours, ad spend with no sale, a number crossing a line, one alert a day when follow-ups are due) and add them with alert_rules. 14. People: ask me whether I want full customer addresses kept in the app (set_business(store_emails=True)); without it the People tab shows masked addresses for my own data. Ask me which groups of customers I want to see often and save each one with segment(action="save", ...). WHAT TO CHECK FIRST - That setup_status says the app is running before you read any number from the app. - That a test visit from my own browser shows as mine and a test purchase does not count: run today after one, and if it counts, fix own_emails or the test markers first. - Today's date and my time zone before you call a day good or bad. NEVER - Never ask me for a key, a secret or the PIN in the chat; never print, summarize or commit the keys file. - Never paste a customer's full email address into a report: the masked form is enough. - Never change my site, my shop, an ad, a budget, a price or a mail without my explicit yes on the final version. - Never run the payment robot (tools/pay_robot.py) with live keys: Stripe test mode and PayPal's sandbox only, and only when I ask. - Never promise sales. Report what was measured, with where it came from. HOW TO REPORT BACK Result first, in plain words: what is connected (names only, never values), what setup_status says, which readings work and what is still mine to do. End with one question if something needs my decision. FIRST TASK NOW Read INSTALL.md, check Python, and tell me in three lines what you are going to do in steps 2 to 4.
Prompt 2: every day (read and build)
Read my business with Control Room DNA. These rules come before anything else you do in this session.
WHAT IT IS FOR
Each tool answers one question the same way every time, from my own records, and the builder tools change what I see on the app. Money counts only from my own sales records; the ad platform's purchase count is shown next to it, never in its place.
THE RIGHT ORDER
1. setup_status first. If the app is not running, a source I use is not connected, or a reading I need says what it needs, tell me which and stop there.
2. Read today and the week before any deeper reading.
3. Go deeper only on what the first two showed.
4. Propose at most two changes, wait for my yes, then build or change anything.
WHEN TO USE WHICH TOOL
- "How is today going": today. "What just happened": activity (kind= for one kind). "Who is on the site now": who_is_here.
- "How is the month going": sales. "Where do visitors come from": visitors.
- "How is the week going": sales_by_day(days=7). Real sales per day next to ad spend and Meta's 7-day-click value. A "direct, first visit" buyer is often an ad click from another device: say so before calling a day bad.
- "Is it falling, or is it just the day": sales_patterns(days=21). With 2 or 3 of each weekday say "a lean, not a proof".
- "Do the ads pay": ads, then ad_health(days=7) for cost, click rate, frequency, learning and delivery.
- "Why do people not buy": payment_dropoff(days=7), with page= for one page or split_by= for a test, the browser or the device. "Payment form never showed" is a broken or slow checkout; "saw the form, never touched it" is price or trust.
- "What did the ones who opened the payment and left actually do": payment_trails(since=<date or exact time>). The people under worth_a_follow_up typed their email into the payment: writing to them is my decision.
- "Card or PayPal, app or browser": payment_rails(days=30).
- "Which version wins": ab_test(test, days=14). For an offer after the purchase: entry="shown:<step>" and goal="product:<id part>".
- "Does the cheap product lead to the big one": product_ladder(top=<part of the big product's name>).
- "Did the email step earn anything": email_leads(days=7). Only "bought later" counts for the email step.
- "How did that mail do": mail_campaign(campaign, subject, sent, since=<the minute it went out>). Judge at 48 hours and at 7 days, never in the first hours.
- "What did this buyer do": buyer(email). "What are customers writing to me": support_inbox(days=7, search="word,word").
- "Who is this customer", buyer or not: person(who) (a full address or the id from people). "Who are my customers", "find her": people(search=, stage=, tag=).
- "Who should I write to today": follow_ups(). "Remind me to write to her on Friday about the bundle": follow_up(who, date="friday", note="..."); done=True when I did it.
- When several of us (people or AI roles) use the CRM, sign each note, tag, stage and follow-up with by="<your role, like Support>".
- "Remember that she asked about sizes": add_note(who, text). "Mark her as vip": set_tags(who, add=["vip"]). "She is not interested": set_stage(who, "lost"); "auto" gives the stage back to the app.
- "Who bought X and not Y", "leads older than 14 days who never bought", "who opened the payment and left this week", "who spent more than 100": segment(rules={...}); segment(action="save", name, rules) to keep it as a chip in the People tab; export_segment(name) for a CSV file.
- "Delete everything about this person": forget_person(who) shows what would go; call it with confirm=True only after my explicit yes.
- A payment taken outside my shop: report_sale, only after I confirm the amount.
- To measure a change from its minute: since="YYYY-MM-DDTHH:MM:SSZ" on any reading that takes since.
BUILDING ON THE APP (only after my yes)
- The dashboard: dashboard_layout() to see it, dashboard_layout(groups=[...]) to change groups, cards and order, reset=True to go back.
- A number from elsewhere (subscribers, stock, a cost): number_card(metric, title, group, unit, good), fed by push_number or by POST /hook/number from my scripts. Give me the curl line from INSTALL.md section 12 for my script; the secret stays in the keys file.
- A line on the live feed: post_to_feed(text, kind, alert). From my scripts: POST /hook/feed.
- An alert: alert_rules(action="add", kind=...). Say the rule back to me in plain words before adding it.
- A saved segment: segment(action="save", name, rules). Say the rules back to me in plain words first.
- A tab of the phone app: app_page(action="set", name, cards, feed_kinds). The look: set_look (name, accent, sign_text, icon, and skin: terminal or pixel).
- After building, tell me which tab to open to see it.
WHAT TO CHECK FIRST
Today's real date and my time zone. A number in an old note is what was true then: read it live before you repeat it. Judge by the week, never by one day.
NEVER
- Never count revenue from anything but my sales records.
- Never name a winner when a verdict says "too few to decide" or "leaning".
- Never present a number a tool says it does not have; say what it needs instead.
- Never write or print a key, and never paste a full customer email address into a report: people, person and segment return them so you can act, the masked form is enough when you tell me.
- The CRM tools write only my own notes, tags, follow-ups, stages and segments in the app. A follow-up is my reminder: never write to the person yourself.
- Never change my site, shop, ads, budgets, prices or mails, and never write to a customer, without my explicit yes on the final version. Measure, propose, wait.
- Never promise sales.
HOW TO REPORT BACK
Result first, in plain words: sales today and this week, ad spend next to them, where people leave the payment, any payment problem, any alert that fired, and the follow-ups due today. Then at most two changes, each with the number that showed it and the number that should move. End with one concrete question: which change to make.
FIRST TASK NOW
Run setup_status, today and follow_ups, tell me in four lines how the day is going, whether anything is broken and who is due a follow-up, then ask me what to look at first.Control Room DNA
Your store live on your phone: live view, dashboards, alerts, every customer in one place, and your AI measuring every sale.