Telegram Channel Scraper – Posts, Search & Monitoring

Pull posts from any public Telegram channel — text, views, dates, forwards, links, and media — as clean CSV/JSON. Then keep it fresh: monitor mode returns only the messages posted since your last run, so a scheduled task becomes a live feed instead of a re-scrape.

No login, no bot token, no phone number. Reads Telegram's public t.me/s/<channel> web preview.

What you can do

Mode	What it does	Fields it uses
Channel posts	Messages from one or more public channels, newest-first.	channels, maxItems
Search	In-channel keyword search across each channel.	channels, searchQueries, maxItems
Channel info	Title, subscribers, description, post/photo/video counts.	channels
Monitor (incremental)	Only messages posted since the last run, per channel. State persists between runs.	channels, maxItems, monitorStoreName

Example input

Latest 100 posts from two channels:

{
  "mode": "channel",
  "channels": ["telegram", "https://t.me/durov"],
  "maxItems": 100
}

Track new posts on a schedule (each run returns only what's new):

{
  "mode": "monitor",
  "channels": ["durov", "telegram"],
  "maxItems": 50
}

Watch crypto channels for announcement keywords:

{
  "mode": "search",
  "channels": ["binance", "coingecko"],
  "searchQueries": ["listing", "partnership"],
  "maxItems": 50
}

@name, name, t.me/name, and https://t.me/s/name all work (case-insensitive).

Scheduled monitoring

To turn monitor mode into a live feed: in the Apify Console go to Schedules → Create, set an interval (e.g. every 15 minutes), and point it at this Actor with mode: "monitor". Each run stores only the messages posted since the last one. Keep the same monitorStoreName across runs so the watermark persists, and give a different name to each separate monitor task. Pipe the dataset to a webhook/Slack/Sheet via Apify integrations to get alerts.

Output (per message)

{
  "type": "message",
  "channel": "telegram",
  "id": 123,
  "url": "https://t.me/telegram/123",
  "text": "…",
  "date": "2026-06-22T15:01:34+00:00",
  "views": 1200000,
  "author": null,
  "isForwarded": false,
  "forwardedFrom": null,
  "tags": ["update"],
  "tagList": "update",
  "mentions": [],
  "links": ["https://…"],
  "firstLink": "https://…",
  "media": [{ "type": "photo", "url": "https://…" }],
  "mediaCount": 1,
  "linkPreview": { "url": "…", "siteName": "…", "title": "…", "description": "…" }
}

tagList, firstLink, and mediaCount are flat mirrors of the array fields, so CSV/Excel exports are usable without unpacking JSON.

What this does — and doesn't — return

To keep expectations honest:

• ✅ Public channel content: text, date, view count, author signature, forwards, replies, hashtags, mentions, outbound links, media URLs, and link-preview cards. Plus in-channel search, channel info, and incremental monitoring.
• ❌ Public channels only — those visible at t.me/s/<channel>. Private channels, groups, and user accounts are not accessible (the Actor rejects invite links with a clear message).
• ❌ No member lists, DMs, poll voters, or per-user reactions. Those require a logged-in account / bot token, which this Actor deliberately does not use. View counts are included; reaction counts are not exposed by the public preview.
• ⚠️ History depth depends on the public preview. The Actor pages backward via ?before=, but coverage is whatever Telegram serves to its web preview — not guaranteed to reach a channel's very first post.

Notes

• Proxy is off by default. The preview pages are public; enable Apify Proxy (or set requestDelayMs to ~300–1000) only for large drains or many channels, to spread the per-IP rate limit.
• maxTotalItems caps results (and charges) across the whole run, independent of the per-channel maxItems.
• One bad/private/typo'd channel is logged and skipped — it won't abort the run or lose the other channels' data.
• How it works / longevity: this reads Telegram's public t.me/s/ HTML preview (no API). That preview has been stable for years, but it is an undocumented endpoint — a Telegram markup change can require a parser update (you'll see a "no posts parsed / markup may have changed" warning), and if Telegram ever added JavaScript bot-protection to it, an HTTP-only scraper like this would need a browser-based rework. No reactions or poll-voter data is exposed by the preview.

Pricing

Pay per result — you're charged once per message (or channel-info record) stored to the dataset. Cap any run with maxTotalItems.