Source Blocks

Available source blocks for ingesting data into workflows

Source Blocks#

Source blocks are the entry points for data into your workflow. They connect to external platforms and pull in content.

RSS Feed Source#

The RSS Feed source monitors any RSS or Atom feed for new content.

Configuration#

FieldDescriptionRequired
Feed URLThe RSS/Atom feed URLYes
Polling IntervalHow often to check (seconds)No
Max ItemsMaximum items per pollNo

Example#

yaml
feed_url: "https://example.com/feed.rss"
polling_interval: 300
max_items: 50

Reddit Source#

The Reddit source monitors subreddits for new posts.

Configuration#

FieldDescriptionRequired
SubredditSubreddit name (without r/)Yes
Sorthot, new, top, risingNo
Time Filterhour, day, week, month, year, allNo
LimitMaximum posts per pollNo

Example#

yaml
subreddit: "netsec"
sort: "new"
time_filter: "day"
limit: 25

Telegram Channel Source#

The Telegram source monitors public Telegram channels.

Configuration#

FieldDescriptionRequired
ChannelChannel username or IDYes
Collection Modeweb (scraping) or agent (API)Yes
AgentRequired for agent modeConditional

Collection Modes#

  • Web Mode: Scrapes public channel content
  • Agent Mode: Uses Telegram API for private channels

Note: Agent mode requires a configured Telegram avatar via the Avatar Selector.

LinkedIn Source#

The LinkedIn source monitors your LinkedIn feed for new posts using browser-based data capture.

Important: This source requires a BYOD Agent with an active LinkedIn avatar. See Configuration Blocks to set up the Avatar Selector.

How It Works#

The agent opens a browser session authenticated with your LinkedIn account, navigates to your feed, and intercepts LinkedIn API responses via XHR capture. Posts are extracted from network traffic — no DOM scraping involved.

Configuration#

FieldDescriptionRequiredDefault
AvatarConnected Avatar Selector with LinkedIn avatarYes
Monitor TypeCurrently only "Timeline"YesTimeline
Capture Wait (seconds)Time to wait for responses (includes scrolling)No60

The capture wait value can be between 10 and 120 seconds. Longer waits collect more posts per cycle but increase execution time.

Data Collected#

FieldDescription
ContentPost text
AuthorAuthor name and profile
TimestampPublication date
MediaImages, videos, shared articles
MetricsLikes, comments, reposts

Requirements#

  1. A deployed BYOD Agent with browser support
  2. An active LinkedIn avatar (authenticated session)
  3. An Avatar Selector node connected to this source

X (Twitter) Source#

The X (Twitter) source monitors tweets using browser-based data capture via your authenticated account.

Important: This source requires a BYOD Agent with an active X/Twitter avatar. See Configuration Blocks to set up the Avatar Selector.

How It Works#

The agent opens a browser session authenticated with your X account and intercepts GraphQL API responses via XHR capture.

Monitor Types#

TypeDescriptionTarget Required
Home TimelineYour personalized feedNo
User TweetsA specific user's tweetsYes (username)
HashtagSearch by hashtagYes (hashtag)

Configuration#

FieldDescriptionRequiredDefault
AvatarConnected Avatar Selector with X avatarYes
Monitor Typetimeline, user_tweets, or hashtagYestimeline
TargetUsername (for User Tweets) or hashtagConditional
Capture Wait (seconds)Time to wait for responsesNo60

Target normalization: Leading @ is stripped from usernames and # from hashtags automatically. For example, @elonmusk becomes elonmusk.

The capture wait value can be between 5 and 120 seconds.

Data Collected#

FieldDescription
ContentFull tweet text
AuthorUsername and display name
TimestampPublication date
MediaPhotos, videos, GIFs
MetricsLikes, retweets, replies, bookmarks

Requirements#

  1. A deployed BYOD Agent with browser support
  2. An active X/Twitter avatar (authenticated session)
  3. An Avatar Selector node connected to this source

Best Practices#

  1. Set reasonable polling intervals to avoid rate limits
  2. Use transforms to filter unwanted content early
  3. Monitor source health from the dashboard
  4. For avatar-based sources (LinkedIn, X, Telegram agent mode), use conservative collection frequencies to avoid account flagging

Next Steps#