Production is down.You’re not at your desk.
Message your own Telegram bot from wherever you are. The agent on your computer opens the metrics in a browser card, reads the logs in a terminal card, finds the commit that broke things — and asks you before it touches production. You answer from the phone.
From “checkout is throwing 500s” to a rollback in ten minutes. Approve the rollback when it asks.
You are on a train when checkout starts failing. You message your bot; the agent reads the dashboard and the logs, finds the commit, asks you on the phone before rolling back, and writes the timeline.
Start it from anywhere
A message to your Telegram bot becomes the agent’s prompt — only from chats you allowed on the card.
Evidence first
The agent reads the real dashboard in a browser card and the real logs in a terminal card before proposing anything.
You approve on the phone
“Needs your input” shows up on your phone through remote access; nothing runs in production until you answer.
The timeline writes itself
Every finding and action goes into an Incident log note, ready for the postmortem.
Ten minutes, four cards
Your phone, your bot, and the machine at your desk.
More ways to ship with agents
Every scenario runs on the cards you already saw — pick the next one.
Keep your on-call agent one message away
It asks before it touches production.
macOS and Linux are coming later.
The first-run wizard installs an agent CLI, Git and the dictation model for you — or skips each step.