Scenario 1: Local use: me and my agents

Who: only the owner. On the other side: the Telegram bots, the Claude Code and Codex sessions, the queues and the timers.

A typical day - 06:00 the Spanish queue publishes a video → "published" event with a link. - 08:00 advviral generates the radar → event with a preview and a "view" button. - 10:00 a request arrives through bot v3 → "request received" event, then "agent X picked it up", then "finished". - 14:00 a Claude session hangs → alert event. - 21:00 daily summary: what came in, what went out and what failed.

What shows on screen - Channels by source: #telegram, #agents, #queues, #publications, #alerts. - Filters by agent, bot and period; highlight only what needs action.

Possibilities - 1a: read-only panel (form A). - 1b: panel where you can reply to a bot or call an agent (form B). - 1c: no new panel, using a private Telegram group as the flow [10].

Open: how much noise is acceptable? Is an agent needed to summarize the flow?

Makes sense when: the problem is "I don't know what is running or what arrived".