Telegram Channel Posts Scraper

Scrape public Telegram channels — posts, view counts, media, forwards, link previews, and exact subscriber counts — as clean, typed JSON. No login, no phone number, no Telegram API key, no session files. Just channel names in, structured data out, at $0.001 per post.

Why this actor?

• No credentials, ever. Works purely from Telegram's public web preview — you never risk an account, share a phone number, or manage MTProto sessions.
• Clean typed output. Numeric views (not "1.72M" strings), ISO 8601 dates, absolute post URLs, typed media arrays, decoded plain text plus original html. Built for LLM pipelines, MCP tools, dashboards, and spreadsheets.
• Honest error handling. A nonexistent name, user account, group chat, or preview-disabled channel produces one free, classified error item (error_code: not_found | group_chat | preview_disabled | ...) — never pages of null junk you paid for.
• Channel metadata included free. Title, description, exact subscriber count, photo, and verified badge ship as a free channel item with every run.
• Deep history. Paginates newest-first to up to 1,000 posts per channel, with an oldestPostDate cutoff for incremental syncs.
• Cheap and fast. HTTP-only (no browser), 256 MB memory, seconds per run, $0.001/post — a fraction of what comparable scrapers charge.

Input

{
  "channels": ["telegram", "@durov", "https://t.me/cointelegraph"],
  "postsLimit": 50,
  "oldestPostDate": "2026-06-01",
  "includeChannelInfo": true
}

• channels — usernames, @-handles, or t.me URLs (mixed freely, deduplicated, max 50).
• postsLimit — newest-first posts per channel (default 20, max 1000). 0 = metadata only.
• oldestPostDate — optional ISO date; stops pagination once older posts are reached.
• includeChannelInfo — emit the free channel-metadata item (default true).

Output

One post item per post (charged), one free channel item per channel, free error items for failed inputs:

{
  "type": "post",
  "input": "durov",
  "channel": "durov",
  "post_id": 531,
  "url": "https://t.me/durov/531",
  "date": "2026-07-16T20:45:25.000Z",
  "text": "Telegram now has a Rich Text Editor...",
  "html": "Telegram now has a <b>Rich Text Editor</b>...",
  "views": 1170000,
  "media": [{ "kind": "video", "url": "https://cdn1.telesco.pe/...", "thumb_url": "https://...", "duration_s": 10 }],
  "link_preview": null,
  "forwarded_from": null,
  "reply_to_post_id": null,
  "author": "Pavel Durov",
  "is_service_message": false,
  "scraped_at": "2026-07-18T19:36:13.707Z"
}

{
  "type": "channel",
  "channel": "telegram",
  "title": "Telegram News",
  "description": "The official Telegram on Telegram...",
  "subscribers": 10049444,
  "photos_count": 13,
  "videos_count": 222,
  "links_count": 367,
  "photo_url": "https://cdn1.telesco.pe/...",
  "verified": true
}

Pricing

Pay per result: one post-scraped event ($0.001) per delivered post. Channel metadata items and error items are always free.

Use cases

• Monitoring & alerting — track news, crypto, and OSINT channels; feed posts into Slack/Discord/email alerts with numeric views for signal ranking.
• LLM & agent pipelines — clean text + metadata for summarization, classification, trend detection, and RAG ingestion (MCP-friendly flat schema).
• Market & competitor research — subscriber growth, posting cadence, engagement (views) across channels, exported to Sheets or a warehouse.
• Archiving — incremental history sync with oldestPostDate + post-id dedupe.

FAQ

Do I need a Telegram account or API key? No. The actor reads Telegram's public web preview. Nothing to register, nothing that can get banned.

Which channels work? Public channels with the web preview enabled (the vast majority of public broadcast channels). Private channels, groups, and user profiles are out of scope by design — you get a classified free error item instead.

Are view counts exact? Telegram publishes abbreviated counts (1.72M). The actor normalizes them to integers (1720000), so they are approximate above 1,000. Subscriber counts on the channel item are exact.

Are member lists or phone numbers scraped? No — this actor deliberately collects no personal data: channel posts and channel-level metadata only.

How fresh are media URLs? Media ships as Telegram CDN URLs, which Telegram time-limits — download promptly if you need the files themselves.