Sources are the watch channels the application collects messages from. These are other people’s channels and chats, RSS feeds and the subscription lists of your account: everything worth tracking, in one place. The collected material goes into the message feed, where it turns into ideas, tags and comments.
Kinds of sources
- RSS/Atom feed — only the feed address is needed;
- Telegram channel through a bot — the channel’s @name;
- Telegram chat (group) through a bot — the chat’s numeric identifier;
- Comment thread through a bot — a numeric identifier;
- Channel through a personal account — the chat’s numeric identifier; this way messages from the channel arrive in real time;
- Folder — combines several sources: a folder gathers everything its child sources collect.
Adding manually
Pick the kind, a name and an address: for RSS — the feed address, for a channel — the @name, for a chat or a comment thread — the numeric identifier. The fetch interval ranges from 1 to 1440 minutes, 15 by default; categories are colored labels, optional. The source appears in the list immediately; the first messages arrive at the next fetch.
Bulk adding from the account’s subscriptions
The “Find in my account” button shows the channels your personal account is subscribed to: title, @name, subscriber count, “owner/admin” labels and search. Check the ones you want — the application creates a source per channel and shows the initial history loading progress; channels that are already sources do not repeat in the list.
The initial history load
The first 50 messages of a source load right away; older ones page in with the “load older” button until Telegram answers that there is nothing newer. Already-known messages are skipped — there will be no duplicates.
Fetching
A background loop checks once a minute whose interval has arrived and pulls the new material — no more than three sources at a time, so Telegram is not overloaded. An unchanged RSS feed is not even re-read: the feed’s answer is checked by its headers. For sources through a personal account, new messages arrive in real time without waiting for the interval, and gaps are patched automatically after a server restart.
No duplicates, protected from failures
Every message is stored once: the reading position is kept at the source and survives restarts. If a fetch fails, the error is recorded and everything already collected, plus the cursor, stays untouched — the next loop continues from the same place.
Categories
Categories are colored labels shared by sources and messages. A message receives its source’s categories automatically; categories can filter the feed and feed into routing.
The sources list
Every source in the list shows: the name, the message and author counts, the interval, the last fetch time with a “fetch now” button, and category chips. When new messages arrive, the counters update in place, without a reload.
The source page
- Overview — the totals: all messages, categories, the interval, the state; for a folder — its child sources;
- Messages — the feed of this source only;
- Authors — who wrote in the source, with message counts and the time of the last post;
- Settings — the name, the interval, a filter expression.
Deleting a source removes everything it collected too — the messages, their tags and links — so the application asks for a confirmation.
What is next
Sorting the collected material is in the Message Feed and Authors sections, automatic triage in Routing, and a polite presence in discussions in AI Comments.