Captures
Overview
Section titled “Overview”Captures panel (left sidebar); mobile Captures tab. The captures pipeline collects screenshots and photos taken via the open-source Orbit Capture SDK (built on Capacitor) and surfaces them inside Orbit for review, annotation, and AI context. Every capture is stored locally and queryable through MCP so AI clients can inspect visual state alongside code.
Capture list
Section titled “Capture list”The Captures panel displays all captures for the current project in reverse-chronological order. Each entry shows a thumbnail, capture timestamp, and any title or notes attached to it.
MCP tool: orbit_list_captures — returns the capture list with metadata (id, timestamp, title, dimensions, notes).
Reading a capture
Section titled “Reading a capture”Open a capture in the panel to view full metadata — dimensions, source, timestamp, notes. The detail view also provides the raw image for visual inspection.
MCP tool: orbit_get_capture — retrieves capture metadata by ID.
Image access
Section titled “Image access”AI clients can fetch the raw image bytes of any capture to include in a multimodal context window.
MCP tool: orbit_get_capture_image — returns the binary image data for a given capture ID.
Updating a capture
Section titled “Updating a capture”Captures support title and notes edits after the fact. Update a capture’s metadata from the detail view or via MCP.
MCP tool: orbit_update_capture — writes new title or notes to a capture record.
Deleting a capture
Section titled “Deleting a capture”Remove a capture from the panel and from local storage. This is permanent.
MCP tool: orbit_delete_capture.
Capture stats
Section titled “Capture stats”Aggregate statistics (total count, storage used, captures by source) are available for the current project.
MCP tool: orbit_capture_stats.
Feedback modal capture
Section titled “Feedback modal capture”The feedback dialog reuses Orbit’s screen-source capture APIs — when you open the feedback form, you can attach a screenshot directly inside the modal. If multiple screens are detected, an inline thumbnail picker lets you select which screen to capture. If screen capture is unavailable, the feedback modal falls back to a Ctrl+V paste slot for a clipboard image.