inflow — a better inbox for LinkedIn Chrome extension security report

inflow is a keyboard-driven inbox for your LinkedIn messages. It runs entirely in your browser, stores everything locally, and is built for people who process a lot of messages and would rather not click through them. Not affiliated with, endorsed by, or associated with LinkedIn or Microsoft. MESSAGING • Send, receive, edit, and unsend messages with file attachments • Optimistic sending with instant UI updates; offline actions queue and replay when you reconnect • Emoji reactions, read receipts, shared-post previews, and draft auto-save • Reply to a specific message, with reply-to indicators and edited-message timestamps • Emoji shortcode autocomplete (:smile) and paste-to-attach for images • New conversation composer with typeahead recipient search INBOX • Four tabs: Focused, Other, Archived, Spam • Star, archive, move to Other, mark read/unread, mark as spam, delete • One-click unread quick-filter and undo for destructive actions • Per-account local storage, so multiple LinkedIn accounts stay separate KEYBOARD-FIRST • J/K to move, Enter to open, E to archive, S to star, U for unread, C to compose • 1/2/3/4 to jump between tabs, / to search, Cmd+K for the command palette • Z to undo, ? for the full shortcut sheet SEARCH • Real-time local filtering across names, messages, and metadata • Server-side LinkedIn search with pagination • Filters: is:unread, is:starred, is:group, has:attachment, has:draft, from:, after:, before:, newer:, older: OPTIONAL AI ASSIST (off by default) • Reply suggestions and inline autocomplete, powered by your own Google Gemini API key • Nothing is sent anywhere until you add a key; remove the key and it stops PRIVACY • No inflow server, no account, no analytics, no telemetry • Your messages are stored only on your own computer, in IndexedDB • Reads your existing LinkedIn browser session — you never enter a password • Full policy: https://github.com/grinich/inflow/blob/main/PRIVACY.md A NOTE ON RISK inflow reads and sends messages through LinkedIn's undocumented internal APIs using your existing browser session. This may be inconsistent with LinkedIn's User Agreement and could put your account at risk of restriction. It is provided as-is, for personal use, at your own risk. Open source (MIT): https://github.com/grinich/inflow