Are you an LLM? Read llms.txt for a summary of the docs, or llms-full.txt for the full context.
Skip to content

How real-time tracking works

Patera keeps a single persistent WebSocket connection to its tracking backend. When a tweet is posted by an account you track, it appears in your dashboard (typically within a second of being published).

What flows over the wire

One socket carries everything the dashboard needs. The three message types you'll mostly care about are:

  • tweet: full tweet payloads as they're posted (text, media, author profile, parent if reply, quote-target if quote).
  • performance: engagement metric updates (likes, reposts, comments, views) batched and pushed live, plus derived velocity and acceleration values.
  • treemap_attention: keyword and topic aggregations across the accounts you track, used to power the AI Summaries panel.

Each message lands in an in-memory store. The live columns re-render against it. Filters and sort are applied client-side, so reordering and re-filtering are instantaneous with no round-trip back to the server.

The same socket also delivers profile-intelligence updates that feed the Profile Trail and the AI Summary pushes that feed AI Summaries. Those pages own the detail; here it's enough to know one connection is the shared transport for all of it.

Clock-skew aware

Tweet timestamps come from the server. Your "X seconds ago" clock comes from your machine. Patera periodically syncs an offset between the two so relative ages stay accurate even if your device clock drifts. You won't see a tweet flash up labelled "in 3 seconds" because of clock drift.

Auto-reconnect and hover-pause

If the connection drops, Patera reconnects automatically the moment it notices, re-opening the socket immediately rather than backing off on a retry timer.

Tweets you already received stay on screen during the drop. The in-memory feed isn't wiped, so nothing you were looking at disappears. Patera does not replay tweets posted while you were offline: the stream picks up live from the instant you reconnect, not from where it left off.

Hover over the Live Feed and incoming tweets queue silently in the background. Your view doesn't reflow while you're reading. Move the mouse away and the queue drains.

There's also a Skip Cached Tweets toggle in the Live Feed options. It drops anything that isn't fresh from the live socket, which is useful when you only want this-second activity rather than back-fill from performance updates.

Focus mode

A Focus mode toggle in the top header hides the Top Performers column and expands the Live Feed to fill the space. AI Actions stays where it is. The setting is remembered across reloads and the button only appears on the dashboard in the default Command Deck layout.

What this means in practice

You're not polling. You're not refreshing. You leave the dashboard open and the world arrives in it.

The same socket feeds Top Performers and the AI Summaries panel. Velocity rankings and trending topics update without any user action.

If the three-column deck isn't the shape you want, the Layout options dropdown in the top header rearranges the dashboard into several persona-tuned layouts (default Command Deck). Your choice is remembered across reloads.