History and Undo
The Recovery History page — every incident and every recovery ever run, the roll call and reminder on each swap, the undo buttons while their window is open, and the report email that closes each recovery.
Everything the program detected and everything it did for you is kept, in order, at Disaster Recovery → History (/recovery/history). It is the same ledger our support team reads before deciding on a grant, so what you see is what they see.
Open incidents
At the top, anything still broken: the kind, the project, the channel or bot it names, the reason Telegram gave, and when it was detected. Each row has a way back into the recovery page with the incident pre-selected. An incident disappears from here the moment a probe finds the thing healthy again or a recovery replaces it.
Recoveries
Below, every operation ever run, newest first, fifteen to a page:
| Column | What it shows |
|---|---|
| Started | When the recovery started — the moment that counts for the 90-day allowance. |
| By | You, Automatic for a failover the platform ran, On Demand for a Swap & Grant of a healthy resource, or Support Grant when a released grant paid for it. |
| Kind | Telegram Account, Project Bot or Channels and Groups, with the project's name. |
| What Changed | Account: Linked to name. Bot: Old Bot replaced by Fresh Bot. Channels: one line per swapped resource with its own Undo or Undone badge, then the roll call. |
| Status | In progress, Completed, Failed or Reverted. |
| Undo until | For a channel recovery still inside its window, the moment the undo closes. |
The roll call on a channel recovery
Under the swapped resources: Re-admitted x / total · Joined y, read live from the invite links — Joined is stamped the moment Telegram approves a member's join request — and, while somebody is still outside and the last reminder is more than an hour old, a Remind button that re-sends the "has moved" message to exactly those members. The same counts and the same button sit on the recovery page while the recovery is running; here they stay after it is done.
Undo
| Recovery | How to undo | Window |
|---|---|---|
| Channel swap | Undo on the resource's line, here or on the recovery page. The resource returns to its previous chat, the links into the replacement are forgotten, everyone is re-admitted to the old chat. | 24 hours |
| Account relink | The signed link in the security email only — it puts the previous account back, signs every browser out, and opens a relink disputed incident for our support team. | 24 hours |
| Bot replacement | No undo. The old bot is gone on Telegram's side and nothing about your members changed. | — |
| Automatic failover | Exactly like a channel swap: Undo on the line, from the page, the history or the link in the failover email. | 24 hours |
An undo spends no allowance. An undone swap keeps its line in the history with the Undone badge, so the record of what happened is never rewritten.
Why the account undo lives in the email
A relink made by someone who took over your dashboard session cannot be defended from that same session. The email reaches the address on your account, not the browser, which is why it is the only place the undo lives.
The report email
Each recovery you run closes with a Recovery complete email: what changed in one or two sentences (the account now linked, the bot replaced, the channels re-pointed with the re-admission counts), a reminder that your plans, subscriptions and payments were not touched, the undo that applies, a button into this page, and the fair-use line. A channel recovery's report waits a quarter of an hour so its counts reflect the queue's work rather than the first second. Automatic failovers have their own email; on-demand swaps, being no recovery, send none.
Example: reading the ledger
Six months in, a creator's history shows:
- 12 Mar · You · Channels and Groups — Signals swapped, Undone the same evening (wrong chat), then swapped again at 21:50; re-admitted 212 / 212 · joined 209. Completed.
- 3 May · Automatic · Channels and Groups — Lounge failed over to its standby at 03:19; re-admitted 340 / 340 · joined 338. Completed. Undo until 4 May 03:19 (passed).
- 3 May · Support Grant · Channels and Groups — the grant support released after the second ban paid for a manual swap of Signals, whose allowance was still spent from March. Completed.
- 20 Jun · On Demand · Channels and Groups — Members moved to a new supergroup; no incident, no allowance. Completed.
- 1 Aug · You · Telegram Account — Linked to Ada (new). Completed.
Support reading the same page sees an account that has been banned twice in the same window, both times swapped back with a clean audit — which is why the grant was released — and a creator who otherwise uses the tools exactly as intended.
Over the API
The same ledger is readable by integrations and the creator apps. GET /v1/recovery/incidents lists what is broken (open by default), GET /v1/recovery/operations lists every recovery with revertible and revert_window_ends_at beside it, and GET /v1/recovery/operations/{operation}/roll-call is the live roll call; the MCP tools list_recovery_incidents, list_recovery_operations and get_recovery_roll_call answer the same questions to an agent. The undo itself stays on this page until the write endpoints land.
How is this guide?