bliptracker

SCOPE ONLINE — bliptracker.xyz

Know which AI chats are waiting on you.

Your AI chats arrive as one flat, endless list — no sense of what's done and what still needs a reply. bliptracker sweeps that list and puts a marker on every title: 🔴 active or ✅ cleared — on desktop and mobile. Your conversations finally get an inbox.

Add to Chrome Get it on GitHub

Free & open source · No account · No server · MIT · bliptracker.xyz

01 — The problem

Your chats have no scope.

So you keep seventeen tabs open as a tracking system — terrified that closing one means losing the thread where the assistant asked you something and is still waiting for an answer.

  • × No way to tell cleared chats from active ones
  • × The chat waiting on you scrolls off the screen
  • × Tabs-as-todo-list goes dark the moment you restart
  • × The running journal you'll never finish shouts as loud as the tax deadline
  • × None of it follows you to your phone
02 — Signal chain

Three steps. The scope does the watching.

1

Teach the chat to transmit

One paste into your AI preferences, once. Installing opens a setup tab that links you straight to the right page, so you're not hunting for it. The prompt asks the assistant to end each substantive reply with a tiny 🔴 / ✅ status line. It appends to whatever you've already got in there, nothing is overwritten, and deleting it later switches the whole thing off.

preference prompt 
End every substantive reply with exactly one
status line as the final line:

🔴 Waiting on you: <what you need from me>
✅ Resolved — safe to archive this chat.

Skip it for trivial exchanges. Never start a
line with 🔴 or ✅ for any other purpose.
2

bliptracker sweeps the signal

From then on, every meaningful answer ends with a status marker — automatically, everywhere you chat, including the mobile app. bliptracker reads the literal signal the assistant wrote.

…so I'd go with the mid-range supplier. Want me to draft the email?

🔴 Waiting on you: confirm the supplier choice

3

The blip lights your list

bliptracker prefixes the chat title with 🔴 (active) or ✅ (cleared). It's exact — it reads what the assistant wrote, it never guesses. Because titles live on the server, the markers show up on your phone too.

Boiler quote
↓
🔴 Boiler quote
03 — On the scope

Small surface. Sharp signal.

Active blips3
  • 🔴 Boiler quote — supplier reply
  • 🔴 Tax return — figures to confirm
  • 🔴 Mortgage broker follow-up

A one-click queue

Click the toolbar icon for a tidy list of every chat waiting on you — each a link straight back into the conversation. The badge keeps the active count in view, so nothing drops off the scope.

A badge you can trust

A count only helps if you believe it. Three red blips you can actually deal with is a to-do list. Three where one's waiting on a delivery and another is a weight log you'll never finish is just noise, and once it's noise you stop opening the popup at all.

Snooze gets the ones you can't do yet out of the way until you can. Pick a time (in an hour, this evening, tomorrow 9am, next Monday, or a date of your own) and it leaves the queue and the badge until then. The title reads 💤🔴 meanwhile, so what's parked is still obvious from your phone. Nothing to remember: it comes back on its own.

Ignore stops the never-ending chats crying wolf. A weight log, a running journal: every entry draws a fresh 🔴, and after a week of false alarms you're skipping the whole list. Ignore it once and it's off the scope for good, marked 🔕 so it stays off on every device. Track again if you change your mind.

Active blips2
  • 🔴 Boiler quote — supplier reply
  • 🔴 Tax return — figures to confirm
💤 Snoozed1
  • Mortgage follow-upwakes Mon 09:00
🔕 Ignored1
  • Weight logoff radar

📱 Syncs to mobile

Markers are part of the title and stored server-side, so they appear in the AI app automatically. Triage the whole scope from the sofa.

⭐ Optional star mirroring

Turn it on and 🔴 active chats get starred — so your Starred section becomes a ready-made queue on every device. Off by default; your manual stars stay yours.

🎯 Exact, never guessed

No fuzzy AI scoring of your chats. bliptracker matches the literal marker the assistant wrote — deterministic and predictable. A blip is a blip.

⚙️ Yours to tune

A simple settings page: master on/off, star mirroring, and how often bliptracker re-sweeps for chats cleared on other devices.

04 — Trust

Your chats never leave your browser.

There's no server to send anything to. bliptracker talks only to your AI provider, using the login you already have — exactly like the website does. No back-channel, no relay, nowhere for a signal to leak.

  • No server. Nothing is uploaded anywhere. There's nowhere for it to go.
  • No analytics, no telemetry. You are not measured or tracked.
  • Local only. Settings and a tiny cache stay on your device.
  • Open source. MIT licensed — read every line before you trust it.
05 — Questions

The fair things to ask.

Which AI assistants does it work with?

Claude, on claude.ai. That's the only adapter that ships today, and bliptracker itself runs on desktop Chrome. The insides are built one adapter per service, so ChatGPT and Gemini are designed and not yet written. Unofficial either way: no affiliation with Anthropic, OpenAI or Google.

Will it overwrite my existing AI preferences?

No. You paste the line in yourself, and it appends to whatever you already have. You stay in full control of your own prompt — that's deliberate.

Is my chat data sent anywhere?

No. There's no backend, no analytics, no third parties. The only network requests go to your AI provider, authenticated by your own browser session — the same calls the site already makes.

Does it work on the mobile app?

The markers do. They're part of the chat title, which is stored server-side, so they appear in the app wherever you see your chat list. bliptracker itself runs on desktop Chrome — it does the sweeping and labelling; your phone just shows the result.

What about chats that never actually finish?

Depends which kind. If you just can't face it yet, snooze it: pick a wake time and it leaves the queue and the badge until then. If it's something that will never resolve, like a weight log or a running journal, ignore it and bliptracker stops classifying it at all. Both are one click in the popup, and both undo. The point is the same either way: the number on the badge should only ever be things you can actually do.

What if the AI site changes and it breaks?

It's possible — bliptracker uses undocumented endpoints. But it's open source, so it can be fixed, and the failure mode is gentle: at worst a marker goes stale, and you can edit any title back by hand. Stand down, nothing destructive.

Is this made by the AI provider?

No. bliptracker is an unofficial community tool, not affiliated with or endorsed by any AI provider. Just something built by someone who wanted their chats to have a scope.

Give your chats a   and a ✅.

Add to Chrome Get it on GitHub

Free & open source · No account · No server · MIT · bliptracker.xyz