Telegram Channel Scraper & Monitor

Scrape public Telegram channels without a login: every post's text, date, views, reactions (each emoji's count, and paid stars), photos, videos, voice notes, files and stickers, polls, link previews, links, hashtags and mentions, and what a post forwards or replies to. Read a channel's history back by date, search it with Telegram's own search, read the comments under each post, and get each channel's exact subscriber count, description and posting pace.

One price, no extras: $1.00 per 1,000 results, with platform usage included and no row cap on Apify's free plan. (Apify's Store page lists platform usage too, because the live API below adds its own; ordinary runs never do.)

It is also a Telegram channel monitor. Schedule it with Only changes and each run returns only what moved since the last one: new posts, edited posts (with the text before the edit) and deleted posts, plus the change in subscribers, posted to Slack, Discord or Microsoft Teams if you like. It runs as a live API and MCP server for AI agents too.

Why this scraper

• No login, no phone number, no session to lose. It reads Telegram's own public web preview (t.me/s/<channel>), the page anyone can open. Scrapers that log in to Telegram break when their account is limited or banned; this one has no account.
• It tracks changes, not just posts. Most Telegram scrapers return the same posts every run. This one keeps a history per channel and returns new posts, edits (what the text said before) and deletions, each once. A post missing from a channel is checked on its own page first: Telegram's "Post not found" reports it deleted at once.
• Exact subscriber counts. The preview rounds them ("9.23M"); each channel row reads the exact count from the channel's own page (9,230,249), and with Only changes, the change since the last run.
• Comments too. Up to 100 comments per post, with the commenter, text, time and what each replies to, from Telegram's own comments widget.
• Search a channel, or its whole history. Telegram's own search finds the newest matches fast; Only posts containing reads the channel back as far as you like and filters it.
• Says what it can't read. A group, a person, a bot, a private invite link, or a channel whose owner turned its preview off gets a free row saying which.

	Typical Telegram channel scrapers	This Actor
Login, phone or session needed	Some (and they break when the account is limited)	Never
New, edited and deleted posts since the last run	You compare runs yourself	Built in (Only changes), with the text before an edit
Alerts to Slack, Discord, Teams	Rare	Yes
Exact subscriber count and its change	Rounded count	Exact, with the change since the last run
Comments	Rare	Up to 100 per post
Reactions per emoji, paid stars, polls, albums	Some	Yes
MCP server for AI agents	Rare	Its own
Price per 1,000 results	About $1 to $10, some plus a fee per run or per channel	$1.00, less on paid plans

Use cases

• News and market monitoring: a schedule every 15 minutes on crypto, finance or news channels that posts each new post to Slack or Discord.
• Brand and keyword watch: Search terms or Only posts containing on the channels of your market, every hour.
• Deleted and edited posts: catch a post a channel edited or deleted, with what it said before (OSINT, journalism, compliance).
• Channel analytics: subscribers, posting pace, average views and reach for a list of channels, tracked over time.
• Research datasets: a channel's whole history back to a date, with views, reactions and media.
• AI agents: ask an assistant "what did @durov post this week, and how many views did it get?" through the MCP server.

How to use it in 4 steps

1. Add Channels: usernames such as durov or @telegram, or links such as https://t.me/s/WatcherGuru. A post link (t.me/durov/400) reads that post.
2. Set Max posts per channel, and Posted after for a date range (2026-10-01 or 7 days).
3. Add Search terms or Only posts containing if you want matches only, and Comments per post if you want comments. Run it.
4. To monitor, switch on Only changes since the last run, add an Alert webhook URL if you want chat alerts, and schedule it.

What you get

One row per post (type: "post"). Example (a real row, an album post of Watcher Guru, 2026-10-27; text and links shortened):

{
    "type": "post",
    "channel": "WatcherGuru",
    "channelTitle": "Watcher Guru",
    "channelId": 1556054753,
    "postId": 15638,
    "url": "https://t.me/WatcherGuru/15638",
    "date": "2026-10-27T15:59:20Z",
    "text": "JUST IN: 🇨🇳 China's tech stock index \"STAR 50\" has now crashed 40% since June as bear market continues. ...",
    "views": 81500,
    "viewsText": "81.5K",
    "reactions": [{"emoji": "❤", "count": 307}, {"emoji": "🤯", "count": 192}, {"emoji": "🔥", "count": 100},
                  {"emoji": "🤣", "count": 75}, {"emoji": "😢", "count": 38}],
    "reactionsTotal": 712,
    "paidStars": 1,
    "media": [{"type": "photo", "url": "https://cdn4.telesco.pe/file/...", "postUrl": "https://t.me/WatcherGuru/15638?single"},
              {"type": "photo", "url": "https://cdn4.telesco.pe/file/...", "postUrl": "https://t.me/WatcherGuru/15639?single"}],
    "albumPostIds": [15638, 15639],
    "mediaTypes": ["photo"],
    "imageUrl": "https://cdn4.telesco.pe/file/...",
    "mentions": ["WatcherGuru"],
    "scrapedAt": "2026-10-28T21:13:59Z"
}

Every post row has its channel, id, link and date, the text with its line breaks, views, reactions, and as it applies: media (photos, videos, GIFs, voice notes, round videos, files, audio, stickers, locations, with links and durations), poll (question, options with their percentages, voters), linkPreview, links, hashtags, mentions, buttons, forwardedFrom, replyTo, author (a signed post), isEdited, and albumPostIds for an album.

A channel row (type: "channel", once per channel a run, with Channel details on): name, subscribers (exact, from the channel's page), description, verified, avatar, the preview's photo, video, link and file counts, the newest post's id and date, postsPerDay, averageViews and viewsPerSubscriber of its newest posts, and with Only changes previousSubscribers and subscribersChange.

A comment row (type: "comment", with Comments per post): the channel and post, commentId, authorName, authorUsername (when public), text, date, replyToCommentId and replyToAuthor, and its link.

Numbers: Telegram's preview rounds views and reactions past 999 to three digits ("34.6K"), so views is that rounding (34,600) and viewsText what the page showed; subscriber counts on channel rows are exact. Media links come from Telegram's servers with an access token and stop working after a while: download what you need soon after the run.

With Only changes, each post row also has change:

change	meaning
newPost	a post published since the last run
edited	the post's text, media or poll changed; previousText is its text before, textChanged says whether the text or only the media changed
deleted	the post left the channel; deletionConfirmed is true when its own page answered "Post not found"
firstRun	the first run's baseline: the newest posts there now

The dataset has views for Posts, Changes, Engagement, Media, Comments, Channels and Not found (a view picks columns: every view lists every row, so filter on type or change if you need to).

How to use it

Channels

A channel is its username (durov, @telegram) or any link to it (t.me/durov, https://t.me/s/durov, telegram.me/durov, tg://resolve?domain=durov). A post link (https://t.me/durov/400) reads that post (and its comments when asked); a photo of an album (?single) reads the album's post. Up to 100 channels and 200 post links a run.

Only public channels have the web preview this Actor reads. A name that is a group, a person or a bot, a private invite link (t.me/+...), a link to a private channel (t.me/c/...), and a channel whose owner turned its preview off get a free notFound row saying which (a channel with its preview off still gets its channel row, with the exact subscriber count). A renamed channel is followed to its new username.

History and dates

Each channel is read newest first, 20 posts a page, back through its history up to Max posts per channel (100 by default, up to 20,000) or to its first post. Posted after stops at the first older post, so 7 days reads only the last week; Posted before reads past newer posts without returning them. Dates are in UTC.

Search terms, keywords and filters

Search terms use Telegram's own search of each channel: fast, and it matches word forms the way Telegram does, but Telegram returns only the newest 20 or so matches per channel and term. For older matches, or several words at once, use Only posts containing: the channel is read back page by page and each post's text, link preview and poll is checked (case and accents don't matter; a hashtag such as #breaking matches the post's hashtag). Reading stops after 500 posts in a row without a match. Exclude words, Only posts with media and Minimum views narrow either. Service posts (pins, a new name or photo, "Channel created") are left out unless you ask for them.

Comments

Comments per post (0 by default, up to 100) reads the newest comments of each post returned, from Telegram's comments widget, as rows after their post; the post row gets commentsCount. Only channels with a discussion group have comments: a post without one gets commentsOff, and after three such posts in a row the channel's comments aren't asked for again. A run reads the comments of up to 500 posts, shared between its channels. Each comment is a result, and Max results counts them.

Monitoring channels on a schedule

1. Switch on Only changes since the last run. Optionally pick the Changes to return.
2. Run it once: the first run returns each channel's newest posts (up to Max posts per channel) as firstRun and saves them as the baseline.
3. Schedule it (Apify Console, Schedules). Each later run returns only the changes.

How it decides:

• Each run reads every post since the last run's newest one, and at least Max posts per channel (the window it watches for edits and deletions), comparing each with the history.
• A post newer than the last run's newest is newPost. An older post the history doesn't know (it was outside the last window) is recorded quietly, not reported.
• A post whose text, media or poll changed is edited, with its text before (up to 600 characters). Views and reactions don't count as edits.
• A post missing from the window is checked on its own page (up to 60 a run): "Post not found" reports it deleted at once; a post whose page still answers is left alone. Without a page check, it is reported after missing two runs in a row.
• Channel searches (Search terms) report new matches only: an edit can move a post out of a search, which isn't a deletion.
• Channel rows carry the subscriber count and its change since the last run.
• Histories are kept per channel, search term and filters (two monitors with different filters don't mix), in the key-value store telegram-monitor-history, or one you name in History name.
• Only what a run returns moves the history: if Max results or your spending limit stops a run, the changes it didn't return come with the next run.

Alerts in Slack, Discord or Microsoft Teams

Paste an incoming-webhook URL in Alert webhook URL (Slack, Discord, Microsoft Teams, or any HTTPS endpoint that takes JSON, such as Zapier, Make or n8n). It switches on Only changes. The first run posts one short "alerts are set up" message; each later run posts its changes (new posts by default, or those you pick in Alert on: edited and deleted posts too), up to 50 per run, each with the channel, text, image, views and link. The URL is treated as a secret and masked in logs. If posting fails, the run still succeeds and its status says why.

Run summary for pipelines

Each run saves a RUN_SUMMARY record in its key-value store: per channel or search, the posts and pages read, the newest and oldest post id read, whether it reached the channel's first post, and the subscriber count; plus the changes, what didn't load and why, the filters used, and the requests made.

Telegram API: use it from code or from an AI assistant

Start a run and get the rows in one call (replace <YOUR_TOKEN>; a call that waits gives up after 5 minutes, so start bigger runs without waiting and read the dataset afterwards):

curl -X POST "https://api.apify.com/v2/acts/akhil_rajesh~telegram-channel-scraper/run-sync-get-dataset-items" \
  -H "Authorization: Bearer <YOUR_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"channels": ["durov", "telegram"], "maxPostsPerChannel": 50, "postedAfter": "30 days"}'

Or from Python with apify-client:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_TOKEN>")
run = client.actor("akhil_rajesh/telegram-channel-scraper").call(run_input={
    "channels": ["WatcherGuru", "cointelegraph"], "keywords": ["bitcoin", "etf"],
    "maxPostsPerChannel": 200, "postedAfter": "14 days", "commentsPerPost": 20,
})
for row in client.dataset(run.default_dataset_id).iterate_items():
    if row["type"] == "post":
        print(row["date"], row["views"], row["text"][:80])

For AI agents: MCP server and live API

The Actor also runs as a live endpoint (Apify Standby): an MCP server that agents call directly, and plain HTTP endpoints. Once a Standby run is up, a request is answered in seconds; the first request starts one. Authenticate with your Apify API token in the Authorization: Bearer <token> header.

MCP server (Streamable HTTP): https://akhil-rajesh--telegram-channel-scraper.apify.actor/mcp

{"mcpServers": {"telegram": {"type": "http", "url": "https://akhil-rajesh--telegram-channel-scraper.apify.actor/mcp",
  "headers": {"Authorization": "Bearer <YOUR_APIFY_TOKEN>"}}}}

In Claude Code:

$claude mcp add --transport http telegram https://akhil-rajesh--telegram-channel-scraper.apify.actor/mcp --header "Authorization: Bearer <YOUR_APIFY_TOKEN>"

A client that can't send headers can use https://akhil-rajesh--telegram-channel-scraper.apify.actor/mcp?token=<YOUR_APIFY_TOKEN>; keep that URL private, as it carries your token. Through Apify's MCP server (https://mcp.apify.com, for Claude, ChatGPT, Cursor and other clients), add the tools with https://mcp.apify.com?tools=akhil_rajesh/telegram-channel-scraper (they appear as akhil_rajesh--telegram-channel-scraper--get_channel_posts and so on), then ask in plain words, for example "what did @durov post this month, and which post got the most views?". Add them that way rather than through Apify's generic call-actor tool, which returns the rows only as text. Apify's MCP server doesn't offer Actors with a live endpoint to agents that pay per call with x402 or Skyfire: connect with an Apify account or API token.

Tool	What it returns
get_channel_posts	Posts of up to 5 channels, newest first, with dates, keywords, excluded words, media and minimum views as filters, and comments if asked
search_channel_posts	Telegram's own search of the channels for your terms (the newest matches)
get_channel_info	One row per channel: exact subscribers, description, verified mark, counters, posting pace and average views
get_posts	Posts by link (t.me/channel/123), with their comments if asked
run_actor	The Actor's whole input in one call

HTTP endpoints, such as GET https://akhil-rajesh--telegram-channel-scraper.apify.actor/posts?channels=durov|telegram&perChannel=20&after=7 days (separate several channels, terms, words or links with |; defaults in brackets). The Actor's Endpoints tab in Apify Console lists them too, and lets you try them.

Endpoint	Parameters
GET /posts	channels, perChannel (20), max (100), after, before, keywords, exclude, media, minViews, service, channelInfo, comments (0)
GET /search	channels, terms, perChannel (22), max (100), channelInfo, comments (0)
GET /channel	channels
GET /post	posts (links), comments (0)
POST /run	The Actor's whole input as the JSON body

Answers carry the rows, rowsByType, notFound (names or posts Telegram can't show: free), notLoaded (what Telegram didn't answer), notRead (what a limit left out: not charged) and notes. A request that reads nothing gets an error and nothing is charged: HTTP 400 for an argument it can't read (a tool error over MCP), 503 when Telegram didn't answer or the run was busy (send it again), 402 when your spending limit for the run is reached (it lifts when the run ends, after 2 minutes without requests). A 503 can also carry rows, when the platform stopped the run while a request was being read: those were stored and charged, and notRead names the rest.

Limits per request: an MCP tool call reads up to 5 channels, 200 rows and 20 comments per post, about 350 kB of rows, within 50 seconds (MCP clients stop waiting at 60 seconds, and Apify's MCP server drops answers over about 1 MB); an HTTP request reads up to 20 channels, 2,000 rows and about 6 MB within 170 seconds. What a limit leaves out is named in notRead and isn't charged. A request that has started runs to its end even if you stop waiting: Apify's Standby proxy keeps it open, so its rows are read and charged. Ask for what you need. Monitoring (Only changes) and alerts belong to scheduled runs: the live endpoint answers each request on its own, so it ignores those fields.

How much does it cost?

You pay $1.00 per 1,000 results (posts, comments and channel rows), less on paid Apify plans, plus a start fee of $0.00005 per run. Platform usage is included in runs of the Actor: there is no compute charge on top, and no row cap on Apify's free plan. (The Store page's pricing box mentions platform usage because of the live endpoint, which charges its own; ordinary runs never do.) Not-found rows are free. With Only changes you pay only for the changes returned (and one channel row per channel, unless you switch Channel details off). The live endpoint (MCP server and HTTP API) charges results at the same prices, plus the Standby run's platform usage while it serves you and until it has been idle for 2 minutes (a short session costs a fraction of a cent).

Apify plan	Per 1,000 results
Free	$1.00
Starter	$0.90
Scale	$0.80
Business and above	$0.70

Run	Results	Cost (Free plan)
A channel's last 1,000 posts	1,001	about $1.00
20 channels, their last 50 posts each	1,020	about $1.02
100 posts with 20 comments each	2,100	about $2.10
A monitor of 10 channels every hour, about 2 new posts a run each	about 720 a day	about $0.72 a day

Apify's free plan ($5 of monthly credit) covers about 5,000 results. To cap spending, lower Max results or set a maximum cost per run in the run options; the Actor stops cleanly at the limit (and with Only changes, the rest comes with the next run).

Good to know

• Speed: a page of 20 posts takes about a third of a second, so 1,000 posts read in about 20 seconds. Comments take one request per post.
• What the preview shows: views and reactions rounded past 999, an "edited" mark without the time of the edit, and no forward counts. Some posts (paid posts, stories, giveaways) aren't shown on the web at all: their row has contentNotShown. A video too big for the web has only its thumbnail (mediaTooBig).
• Media links expire: they carry Telegram's access token. Download what you need soon after the run.
• Refusals: a run that meets a refusal (Telegram's "Too Many Requests") waits and asks again, then names what it couldn't read and charges nothing for it.
• Groups and private chats need a Telegram account to read, which this Actor never uses.

FAQ

Do I need a Telegram account, API id or phone number? No. The Actor reads the public web preview of channels, which needs no login.

Can I get alerts for new posts in a Telegram channel? Yes: switch on Only changes, paste a Slack, Discord or Teams webhook in Alert webhook URL, and schedule it (every 5 to 60 minutes). Each new post is posted once.

Can it detect deleted or edited Telegram posts? Yes, with Only changes: an edited post comes with its text before the edit, and a deleted post is confirmed on its own page ("Post not found") before it is reported.

Can I scrape a channel's whole history? Yes: set Max posts per channel high (up to 20,000) or Posted after to a date, and it reads back page by page to the channel's first post.

Can I search Telegram for a keyword? Within the channels you name, yes: Search terms (Telegram's own search, newest matches) or Only posts containing (the whole history, filtered). A search across all of Telegram needs a logged-in account, which this Actor doesn't use.

Can it scrape Telegram groups or members? No: groups and their members need a login. A group's name gets a free row saying so.

Does it download photos and videos? It returns their links (and thumbnails, durations and file names); the links work for a while, so download them soon after the run.

Are view counts exact? Telegram's preview rounds them past 999 ("34.6K"), and so does this Actor's views. Subscriber counts on channel rows are exact.

Do I need a proxy? No: Telegram answers Apify's servers directly. You can use your own proxies or Apify's datacenter proxies; residential proxies aren't available in this Actor, as they're billed per GB on top of the result price.

Is it legal? The Actor reads public channel pages that Telegram shows to anyone on the web, at a modest pace, without logging in. Use the data in line with Telegram's terms and the laws that apply to you, especially for personal data such as commenters' names.

Other Actors by this developer

• Google Ads Transparency API & Scraper: every ad a brand runs on Google Search, YouTube, Shopping and Maps, with ad copy and targeting, political ad spend and ads Google removed, plus Slack, Discord or Teams alerts for new and stopped ads
• Craigslist Scraper & Tracker: Craigslist listings by keyword, city or state, with price drops, new and gone listings, and alerts
• Kijiji Scraper & Tracker: Kijiji listings by keyword, city and category with prices and price drops, cars and rentals, plus Slack, Discord or Teams alerts
• Yahoo Finance Scraper & API: stock quotes, history, financials, options and news, with an MCP server
• App Store Reviews Scraper & API: App Store reviews, ratings and rankings, with Slack, Discord or Teams alerts for new reviews

Disclaimer

This Actor is an independent tool. It is not affiliated with, endorsed by or sponsored by Telegram. It reads publicly available channel pages at a modest pace and never logs in. Use the data in line with Telegram's terms and the laws that apply to you.