Skip to content

High-performance browser automation bridge and multi-instance orchestrator with advanced stealth injection and real-time dashboard.

docs: GET /visible and the capture snapshot's visible field answer different questions under the same name (PIN-188)

Luigi Agosti7f3303095fdf
diff --git a/docs/endpoints.md b/docs/endpoints.md
index ae16155e..0c3ed0d5 100644
--- a/docs/endpoints.md
+++ b/docs/endpoints.md
@@ -153,6 +153,8 @@ GET /snapshot
GET /tabs/{id}/snapshot
GET /text
GET /tabs/{id}/text
+GET /visible
+GET /tabs/{id}/visible
POST /find
POST /tabs/{id}/find
POST /evaluate
@@ -163,6 +165,8 @@ POST /tabs/{id}/evaluate
`GET /action` decodes a subset of the action fields and refuses, with `400` naming the field, any parameter it cannot express rather than silently dropping it — so a modifier chord, a drag, `waitNav` or `humanize` must be sent as `POST /action` with a JSON body.
+`GET /visible` (and `pinchtab visible <ref>`) answers CSS rendered-ness — `display`, `visibility`, `opacity`, and a laid-out box with non-zero size. Scroll position is not an input: an element far below the fold, or scrolled past, still reports `visible: true`. On-screen-ness is the response's `onScreen` field, which shares the capture snapshot's viewport-intersection predicate (see [reference/capture.md](reference/capture.md)); `onScreen` is omitted when the element could not be measured — absent means unknown, never "no".
+
Action kinds currently include:
- `click`