Skip to content

Domain Watch

A typical capture is mostly noise. Open one app and you'll see requests to analytics endpoints, CDN edges, push-notification keepalives, font hosts, and a dozen other domains that have nothing to do with the bug you're chasing. Domain Watch filters that out at capture time, so the log shows the API you actually care about.

The rule is simple: watched domains capture every request normally. Unwatched domains show only the first request as a teaser, and subsequent requests to the same domain are dropped before they reach the log.

When traffic from a new (unwatched) domain hits Probe, the first request lands in the log as a normal row — that's the teaser. Click it, and the Detail panel is replaced by a Domain Summary panel: host name, request count seen so far, and a Watch Domain button.

Domain Summary panel for an unwatched domain: a globe icon, the host "sentry.io", a "1 request · 1 subdomain" line, a per-method count, and a "Watch Domain" button

If you don't watch the domain, every subsequent request to it is silently dropped. They never enter the recording log, so they don't bloat the captured session, don't slow down the UI, and don't show up in Sessions you save later.

If you click Watch Domain, future requests to that host capture in full. Earlier dropped requests aren't recovered — they were never recorded in the first place.

There are three ways to flip the toggle:

  • Right-click a domain in the sidebar tree → Watch or Unwatch. The fastest path when the domain is already visible.
  • Click the teaser, then Watch Domain in the summary panel.
  • View → Watched Domains menu (Cmd+Shift+D on macOS) → toggle the row.

The toolbar shows a teal eye icon when at least one domain is watched.

Open it from View → Watched Domains (Cmd+Shift+D on macOS). The dialog lists the root domains you've watched — including ones you've switched off — not every domain Probe has seen. It lets you:

  • Toggle watch on / off per domain. Off keeps the entry in the list but treats the domain as unwatched.
  • Delete a domain from the list. It goes back to unwatched; requests to it are still dropped until you clear the log or start a new recording tab, after which the next request is a fresh teaser.
  • Clear all to remove every entry at once (every domain goes back to unwatched).

Watch state persists across app restarts, so domains you watched yesterday stay watched today. The list of dropped/teasered domains is per-session.

The pattern that pays off: when you sit down to debug something, you usually know which domain matters. Visit your app, watch the one or two domains you care about, and from then on the log is clean — your endpoints in full detail, everything else condensed to a single teaser row per host.

Concrete examples:

  • Debugging the staging API for a mobile app — watch api.staging.example.com. Telemetry, fonts, image CDN, push servers all collapse to teasers.
  • Auditing a third-party SDK's traffic — watch only that SDK's domain to confirm what it sends home.
  • A long QA session — start the day, watch your two backend hosts, capture for hours without filling memory with noise.

Domain Watch's quieter sibling. Right-click any domain in the sidebar and choose Ignore to add it to the Ignored list — Probe drops every future request to that host before it reaches the log, the same way unwatched domains drop after the teaser, but without ever capturing a teaser row. Ignored hosts persist across app restarts. Click Ignored in the sidebar to open the manager, where you can review the list, click the Ignore Domain button to add a host by name, and remove an entry to start capturing it again.

The same list also holds ignored clients: apps (a process name such as Slack) or a phone's IP address. Probe forwards a client's traffic untouched, the same as an ignored host. The easiest way to add one is Hide this app in the sidebar's Apps menu. A first-run default list is already in place: well-known Apple, Google, Microsoft and Meta hosts, crash-reporting and analytics services, and the chat apps Slack, Discord, LINE, Telegram, Zalo and WeChat. Remove any entry to capture it again; a newer version of Probe can add new defaults once.

Use Ignore for domains you actively know you don't want — your team's metrics endpoint, OS push keepalives, telemetry that never matters. Use Domain Watch for the inverse case: you don't know which domains matter yet, so you let teasers in and watch the ones you care about.

  • Domain Watch decides what gets captured. Dropped traffic is gone.
  • Filtered Lists decide what gets shown. Filtering is reversible — change the filter, the rows reappear.

Use Domain Watch to keep the log lean from the start. Use Filtered Lists to slice and dice traffic that's already captured.