High-performance browser automation bridge and multi-instance orchestrator with advanced stealth injection and real-time dashboard.
Frame
Get or set the current frame scope for selector-based snapshots and actions.
By default, selector lookup stays in the main document. To target iframe content with CSS, XPath, or text selectors, set the frame first.
Refs from /snapshot are different: if a snapshot includes same-origin iframe descendants, those refs can still be used directly without setting frame scope.
curl http://localhost:9867/frame
curl -X POST http://localhost:9867/frame \
-H "Content-Type: application/json" \
-d '{"target":"#payment-frame"}'
curl -X POST http://localhost:9867/frame \
-H "Content-Type: application/json" \
-d '{"target":"main"}'
# CLI Alternative
pinchtab frame # Shows: main (or frameId if scoped)
pinchtab frame "#payment-frame" # Shows: <frameId> (<name>)
pinchtab frame main # Shows: main
pinchtab frame --json # Full JSON response
Targets accepted by POST /frame and pinchtab frame:
mainto clear frame scope- a snapshot ref for an iframe owner
- a selector for an iframe element
- a frame name or frame URL
Response: {tabId, scoped, target, current} — unscoped, target and current
are "main"; scoped, target is the frame ID and current/frame carry
{frameId, frameUrl, frameName, ownerRef}. POST without target returns 400;
a target that is not an iframe or frame returns 400.
MCP: pinchtab_frame takes target (omit to read the current scope), tabId,
browser.
Typical iframe flow (API form exercised in tests/e2e/scenarios/api/actions-extended.sh):
pinchtab snap -i
pinchtab frame "#payment-frame"
pinchtab snap -i
pinchtab fill "#card-number" "4111111111111111"
pinchtab click "#pay-button"
pinchtab frame main
Notes:
- selector scope is explicit; unscoped selectors do not automatically pierce into iframes
- same-origin iframe content is supported; cross-origin iframe descendants are not currently exposed as frame scopes
- nested iframes usually require multiple
framehops - the frame scope applies to
/snapshot,/capture, selector-based/actioncalls, and/textwhenframeIdis not provided explicitly - a read served from a frame scope carries a
frameobject (frameId,frameUrl,frameName,ownerRef,frameTitle) so a later reader can tell the content is not the top document /evaluateis separate and does not inherit frame scope