Fetch it. Find it. Use it.

Start with a question: “Where was that trip mentioned?” or “Which chat was it in?” Pipe indexit into an interactive tool and get straight to the answer.

Sign in first. The examples use public placeholder names; substitute your own channel or group. The browser illustrations below use fictional data and do not connect to Telegram.

1. Explore messages with fx

Install fx (brew install fx on macOS), then run:


indexit -q telegram fetch messages \
  --dialog @example_channel --limit 200 | fx

Open a message, inspect its sender, and find a word in its text. In the browser example, try trip, then @ from.display and click a matching field.

fx / explore messagesTRY IT HERE

$ indexit -q telegram fetch messages \
  --dialog @example_channel --limit 200 | fx

Try a search

[0] {5}

from {2}

media {1}

[1] {4}

from {2}

[2] {4}

from {2}

Explore a record, or search across its fields.

Simplified browser illustration · fictional data · no Telegram connection

In the real terminal, use these fx controls :

Action	Key
Navigate and expand records	Arrow keys
Search values with a regular expression	`/`, type your query, Enter
Move between text matches	`n` / `N`
Fuzzy-search JSON paths	`@`
Print the selected value to stdout	`P`
Show keyboard help	`?`

Path search finds fields such as from.display; content search finds words inside messages. The browser illustration uses literal text matching and a simple fuzzy path matcher; fx provides the full terminal behavior.

Use the native fx binary (Homebrew or a release download) for this workflow. The npm package only provides non-interactive processing.

2. Find a conversation with fzf

Install jq and fzf . On macOS: brew install jq fzf.


indexit -q telegram fetch dialogs \
  | jq --unbuffered -r '[.uid, .title // .username // .uid] | @tsv' \
  | fzf --delimiter '\t' --with-nth 2..

Type a few letters from the title, move with ↑ / ↓, and press Enter. In the browser example, wknd finds Weekend trips even though the letters are not adjacent.

fzf / find a conversationTRY IT HERE

$ indexit -q telegram fetch dialogs \
  | jq --unbuffered -r '[.uid, .title] | @tsv' \
  | fzf --delimiter '\t' --with-nth 2..

Try a search

6 / 6 conversations

Type a few letters, then choose a conversation.

Simplified browser illustration · fictional data · no Telegram connection

jq turns each dialog into a row containing its UID and title. fzf displays the title and returns the original row on selection, so you keep the address for the next command. Press Esc to cancel.

--unbuffered lets rows reach fzf as they arrive. This example illustrates fuzzy matching; fzf also supports exact matches and exclusions .

Search messages as a list

The same pattern works when you remember a phrase but not the message:


indexit -q telegram fetch messages \
  --dialog https://t.me/example_channel --limit 200 \
  | jq --unbuffered -r '[.id, .date, .text] | @tsv' \
  | fzf --delimiter '\t' --with-nth 2..

Each message stays on one row, including messages with line breaks. Selecting a row returns its message ID, date, and text. Use that ID with fetch message to inspect the complete record.

Keep the first look small

Both workflows search the data fetched in this invocation. They do not search all of Telegram. --limit 200 means the latest 200 emitted messages; increase it or choose a date window when needed. fx opens the fetched input for exploration; fzf can update its candidate list while records arrive.

The -q flag suppresses progress chatter, not errors. All three tools run locally and receive data through stdout — no intermediate export file is needed.