Architecture and data flow

Runtime components

Background

src/background/index.ts is the coordinator:

• initializes settings on install
• reacts to bookmark creation events
• also reacts to bookmark removal, changes, and moves to keep the dashboard fresh
• handles runtime messages
• accepts review items and triggers reruns
• removes accepted or dismissed queue items instead of keeping them around

Content script

src/content/index.ts extracts page title, meta description, excerpt, and optional full text from an open tab.

If no matching open tab can be queried, the pipeline falls back to fetching the bookmark URL.

UI

• popup for quick status and reruns
• options page for the full dashboard
• dashboard widgets and modals for settings, root filters, delete actions, queue handling, and paginated event logs with AI details

Pipeline flow

src/lib/pipeline.ts performs:

1. settings load
2. bookmark lookup
3. structure snapshot load
4. exclusion filtering
5. preferred root derivation from the current bookmark path
6. page snapshot extraction
7. provider classification
8. move or queue handling

Outcomes

• moved
• queued
• skipped
• failed

Runtime messages

Message	Purpose
dashboard:get	load dashboard state
settings:update	save settings
bookmark:run	process one bookmark manually
bookmark:delete	delete one bookmark
queue:accept	apply a queued decision
queue:dismiss	dismiss a queue item
queue:rerun	rerun one queue item
popup:current-tab	inspect the active tab
options:open	open the options page
content:extract	read page data from the content script

Storage

src/lib/storage.ts manages:

• bookmark-sorter-settings
• bookmark-sorter-review-queue
• bookmark-sorter-event-log
• bookmark-sorter-dashboard-revision

Recent events are stored as structured EventLogEntry objects, including optional AI request and response payloads.