Youtube Live Chat Scraper

Extract live chat messages from YouTube streams and replays — messages, Super Chats, membership events, badges, timestamps. Replay-first, export-ready for moderation and analytics.

Provide video URLs or channel handles and get a clean, structured dataset of every chat event. No browser required.

What does YouTube Live Chat Scraper do?

YouTube Live Chat Scraper is an Apify Actor that extracts live chat data from YouTube live streams and live replays using HTTP requests — no browser needed. It captures regular messages, Super Chats, Super Stickers, membership events, and moderation actions, then outputs them as a clean, analytics-ready dataset.

Key capabilities

• Replay chat extraction — Extract the full chat replay from any YouTube live stream with chat replay enabled
• Live chat capture — Best-effort real-time polling of live chat during active streams
• Channel discovery — Provide a channel URL or @handle and auto-discover live streams and recent replays
• Super Chats & Stickers — Capture paid messages with amounts, currencies, and tier colors
• Membership events — Track member joins and renewals shown in chat
• Moderation events — Capture message deletions, user bans, and system events
• Batch processing — Process multiple videos or channels in parallel
• Normalized schema — Stable output format with consistent event types across all renderers

Use cases

• Moderation analytics — Audit chat for spam, toxicity patterns, and moderation effectiveness
• Audience research — Analyze engagement peaks, sentiment, and discussion topics during live events
• Revenue tracking — Collect Super Chat amounts and donor messages for creator analytics
• Campaign measurement — Measure audience reaction during sponsored streams or product launches
• AI/NLP datasets — Build training data for sentiment analysis, topic modeling, or spam classifiers
• Community monitoring — Track live chat activity across multiple channels automatically

Input

You can provide video URLs/IDs directly, channel URLs/@handles for auto-discovery, or both. At least one of videoRefs or channelRefs is required.

Input example — Replay extraction (direct video)

{
  "videoRefs": [
    "https://www.youtube.com/watch?v=2VPo6OHrltU"
  ],
  "mode": "replay",
  "maxMessagesPerVideo": 5000,
  "maxTimePerVideoSecs": 600,
  "includeSuperChats": true,
  "includeMembershipEvents": true,
  "includeModerationEvents": true
}

Input example — Channel discovery

{
  "channelRefs": [
    "https://www.youtube.com/@TheCypherbyDbloc"
  ],
  "channelDiscoveryMode": "current_live_then_replays",
  "maxVideosPerChannel": 5,
  "lookbackDays": 30,
  "mode": "replay",
  "maxMessagesPerVideo": 5000,
  "maxTimePerVideoSecs": 600
}

Input parameters

Videos

Field	Type	Default	Description
videoRefs	string[]	[]	YouTube watch URLs or 11-char video IDs
mode	string	"replay"	replay or live

Channel Discovery

Field	Type	Default	Description
channelRefs	string[]	[]	YouTube channel URLs, @handles, or channel IDs (UC...)
channelDiscoveryMode	string	"current_live_then_replays"	current_live_only, latest_replays_only, or current_live_then_replays
maxVideosPerChannel	integer	5	Max videos to discover per channel (1–50)
lookbackDays	integer	30	Only discover replays from the last N days (0 = no limit)
includeUpcomingLives	boolean	false	Include scheduled/upcoming live streams
includeMembersOnly	boolean	false	Include members-only streams

Extraction limits

Field	Type	Default	Description
maxMessagesPerVideo	integer	5000	Max messages per video (1–200,000)
maxTimePerVideoSecs	integer	600	Time budget per video in seconds

Event filters

Field	Type	Default	Description
includeSuperChats	boolean	true	Extract Super Chats and Super Stickers
includeMembershipEvents	boolean	true	Track membership events
includeModerationEvents	boolean	true	Track moderation events

Live mode controls

Field	Type	Default	Description
liveCaptureDurationSecs	integer	300	Total capture duration for live mode
pollIntervalMs	integer	2000	Poll interval in milliseconds (1000–10000)
stopWhenStreamEnds	boolean	true	Stop if stream ended is detected

Localization

Field	Type	Default	Description
language	string	"en"	Localization language code
regionCode	string	"US"	Localization region code

Advanced settings

Field	Type	Default	Description
maxConcurrency	integer	3	Parallel video processing (1–10)
useProxy	boolean	false	Use Apify residential proxy
proxyCountryCode	string	""	Proxy exit country
requestTimeoutSecs	integer	30	HTTP timeout per request
maxRetries	integer	3	Retries for 429/5xx errors
minDelayMs	integer	0	Delay between continuation requests
debug	boolean	false	Enable verbose logging

Output

The actor outputs one dataset row per chat event. Each row contains the message content, author information, event metadata, and discovery provenance.

Event types: message, super_chat, super_sticker, membership, moderation, system, error

Output example (JSON)

{
  "videoId": "BZ_UzbpMjMg",
  "videoUrl": "https://www.youtube.com/watch?v=BZ_UzbpMjMg",
  "eventType": "message",
  "messageId": "ChwKGkNNamR6OTJMbEpNREZkSUwxZ0FkaVRRYVpR",
  "timestampUsec": 1773103087877500,
  "publishedAt": "2026-03-10T00:38:07.877Z",
  "authorChannelId": "UC9N4Cb1knbL1DJE1K1S7xnQ",
  "authorName": null,
  "authorHandle": "@kasparaskriauciunas2433",
  "authorPhotoUrl": "https://yt4.ggpht.com/ytc/AIdro_l0N2E6QEsxEQF9oCJnhf1j9wAfU22I374hUQAbatg=s32-c-k-c0x00ffffff-no-rj",
  "authorBadges": [],
  "messageText": "@ByornVarg what if we dont have enough gas, whats ur thoughts then?",
  "messageParts": [
    {
      "type": "text",
      "text": "@ByornVarg what if we dont have enough gas, whats ur thoughts then?"
    }
  ],
  "isPinned": null,
  "likeCount": null,
  "superChatAmount": null,
  "superChatCurrency": null,
  "superChatColor": null,
  "membershipLevel": null,
  "moderationAction": null,
  "seq": 1,
  "regionCode": "US",
  "retrievedAt": "2026-03-10T01:24:20.274Z",
  "streamStatus": "replay",
  "runId": "qg6A5QXp9LNumBfpA",
  "discoverySource": "channel_discovery_api",
  "channelRef": "https://www.youtube.com/@TheCypherbyDbloc",
  "discoveredAt": "2026-03-10T01:24:19.503Z",
  "isDiscoveredLive": true,
  "isDiscoveredUpcoming": false
}

Output screenshot — All fields table view

Output fields

Field	Type	Description
retrievedAt	string (ISO)	When the event was captured
videoId	string	YouTube video ID
videoUrl	string	Canonical watch URL
streamStatus	string	live, replay, ended, or unknown
eventType	string	message, super_chat, super_sticker, membership, moderation, system, error
messageId	string	Stable message ID
seq	integer	Sequential index per video (1..N)
publishedAt	string (ISO)	Message timestamp (when available)
timestampUsec	integer	Raw microsecond timestamp
authorChannelId	string	Author's channel ID
authorName	string	Display name
authorHandle	string	@handle
authorPhotoUrl	string	Avatar URL
authorBadges	string[]	e.g., ["OWNER", "MODERATOR", "MEMBER"]
messageText	string	Chat message text
messageParts	object[]	Structured runs (text + emojis)
isPinned	boolean	Whether message is pinned
likeCount	integer	Like count (rare)
superChatAmount	string	e.g., "$5.00"
superChatCurrency	string	e.g., "USD"
superChatColor	string	UI tier color
membershipLevel	string	e.g., "Silver Member"
moderationAction	string	e.g., "MESSAGE_DELETED"
regionCode	string	Region used for request
runId	string	Apify run ID
discoverySource	string	direct_videoRef, channel_discovery_api, or channel_discovery_html
channelRef	string	Channel input that produced this video
discoveredAt	string (ISO)	When the video was discovered
isDiscoveredLive	boolean	Whether video was live when discovered
isDiscoveredUpcoming	boolean	Whether video was upcoming when discovered

Run summary

The actor writes a RUN_SUMMARY.json to the key-value store with per-video totals, errors, and channel discovery results.

How it works

1. Channel discovery (if channelRefs provided) — Resolves channel handles/URLs to channel IDs, discovers live streams and replays using YouTube Data API with HTML fallback, and applies discovery filters
2. Token extraction — Fetches YouTube watch page HTML and extracts innertube API key, context, and chat continuation token
3. Chat pagination — Paginates chat events using continuation tokens via the youtubei API (replay) or polls live chat (live mode)
4. Event normalization — Maps YouTube renderer types into stable event types (message, super_chat, etc.)
5. Output — Pushes clean dataset rows and writes RUN_SUMMARY.json

Performance

• Replay mode — 5,000 messages typically in 1–5 minutes
• Live mode — Near real-time at 1–10 seconds latency
• Channel discovery — Adds ~2–5 seconds per channel
• Concurrency — Up to 10 videos processed in parallel
• HTTP-only — No browser overhead

Limitations

• Chat replay must be enabled on the video
• Members-only chat behind authentication is best-effort
• Live mode is best-effort polling (not websocket)
• Age-gated or restricted content may not be accessible
• Channel discovery depends on YouTube Data API quota and/or streams tab availability

Pricing

This actor uses a pay-per-event pricing model. You are charged for each chat event (dataset row) extracted.

Event	Description
chat-event	1 event per dataset row