Text hotkeys

I would like you to implement it

/loop I would like you to proceed to run it, see what needs to be improved and iteratively develop and run it until it is fixed for it

I would like you to look through the output and see what may be improved for it

I would like you to deploy it as a new instance container, overriding the old one and updating it

Run an exhaustive check of every communication channel and tell me what needs a reply. FIRST health-check each channel before reporting anything, because a dead channel looks identical to a quiet one: for each channel (the five mailboxes, SMS, WhatsApp, LinkedIn) state the timestamp of the last message actually ingested and whether the ingest path is alive (task/process running, endpoint reachable, token valid), and flag anything stale or unverifiable as DEGRADED or DEAD rather than reporting 'nothing pending'. Then sweep all five mailboxes (levitinvlad@hotmail.com, bergvictor1999@outlook.com, levitinvlad99@gmail.com, bergvictor1999@gmail.com, victorberg12061999@gmail.com) across EVERY folder — Inbox, Junk/Spam, Deleted/Trash, Archive, Sent, Drafts, every label and subfolder, plus the [SMS] and [WA] folders — both read and unread, over the last 7 weeks anchored on today, paging through every message without truncating or sampling. Also cover LinkedIn DMs, SMS, WhatsApp, and my calendar / upcoming meetings. Dedupe the SMS and WA folders (they are mirrored across hotmail and outlook, and SMS forwards arrive FROM my own address, so do not let an own-sender filter drop them). Do not skip Norwegian or Russian mail — the government, bank and consulate deadlines are there. Include every human message with a pending ask or awaiting my response (employer, recruiters, invoices, claims, kommune, NAV, Skatteetaten, tribunals, parking, consulate, personal), plus account-integrity events such as new SIM or eSIM orders, new-device logins and username or password changes even when nobody is asking me anything. Before flagging anything, verify I have not already replied: check Sent AND Drafts across ALL accounts, not just the receiving one, matching on thread and normalised subject — my LinkedIn replies go out via the hit-reply@linkedin.com relay and show up in outlook Sent Items as 'RE: Message replied: ...', and a written-but-unsent draft is NOT a reply. Never propose a duplicate reply. Ignore newsletters, automated notifications, job alerts, mass-application receipts and 2FA codes unless they carry a real action or deadline. Report anything due within 72 hours or already overdue FIRST, then group into: needs a reply from me (sender, what they want, deadline, recommended action, how long it has been waiting) / waiting on them / already replied / channel health with per-account and per-folder message counts so the coverage is auditable. Then offer to draft the replies, but do not send anything outward without showing me the exact final wording and getting my explicit approval.

Quick comms triage: sweep all five mailboxes (levitinvlad@hotmail.com, bergvictor1999@outlook.com, levitinvlad99@gmail.com, bergvictor1999@gmail.com, victorberg12061999@gmail.com) plus SMS, WhatsApp, LinkedIn DMs and my calendar over the last 7 days only — Inbox, Junk and the [SMS]/[WA] folders, read and unread. First say whether each channel is actually alive (last message ingested), so a broken channel is never reported as 'nothing new'. Then show me only what is genuinely time-critical: anything due within 72 hours or already overdue, anything a human is waiting on me for, and any account-integrity event (new SIM/eSIM, new-device login, password or username change). Check Sent and Drafts across all accounts before flagging, so you never propose a duplicate reply. Skip newsletters, job alerts and application receipts entirely. Keep it short — one prioritised list, no full inventory. Offer to draft replies but send nothing without my explicit approval of the exact wording.