PostMarX

Better bookmarks for X, kept local in your browser.

PostMarX is an independent project and is not affiliated with X Corp. or Google. PostMarX is a local-first bookmark manager for X. Tap the bookmark button on a post and PostMarX stores it in this Chrome profile. Search by words or meaning, file posts into categories, label them with tags, and import bookmarks and folders already saved on X. There is no PostMarX account, hosted backend, telemetry, or cross-device sync. The included meaning-search model and WebAssembly runtime ship inside the extension and run in Chrome. Settings also offers reviewed optional models. If you uninstall the extension or remove this Chrome profile, the archive is deleted unless you exported a JSON backup. **What you can do** - Save only to PostMarX, or keep PostMarX and X bookmarks in sync - Search immediately by words and optionally by meaning on this device - Preview saved photos and open play-marked video thumbnails on X - Organize with categories, tags, automatic-tag previews, and bulk actions - Import X bookmarks and folders with visible progress, cancel, and safe retry - Export and restore PostMarX JSON backups - Choose an included or downloaded local meaning-search model **What stays local** Saved posts, labels, settings, and embedding vectors remain in this browser. No PostMarX server receives the archive. PostMarX does not sell or share this data and includes no analytics, telemetry, or crash reporting. Optional model downloads start only after user action. Hugging Face receives ordinary download metadata, including the user's IP address and request headers. Bookmark and search text are not sent to Hugging Face. Downloads are pinned model data, not executable code. The included model requires no download. Optional model files are not included in JSON backup exports. No PostMarX server receives archive or model traffic. When an archive view shows a saved avatar, photo, or video thumbnail, Chrome can request it directly from X's approved image host, `pbs.twimg.com`. X can receive the user's IP address and ordinary request metadata for that image request. No PostMarX server receives the archive or proxies the image. PostMarX drops unapproved avatar/media origins and checks stored preview URLs again before rendering them. **Why PostMarX accesses X** Manifest host access is limited to x.com and twitter.com for the visible capture and import features. The extension-page image policy separately allows saved previews only from `https://pbs.twimg.com`. Assisted import reads the X bookmark pages. Folder discovery can observe and make same-origin X GraphQL requests using transient page authentication and CSRF context. That context remains ephemeral in the X page and is not stored or sent to PostMarX. To adapt to X's rotating folder-operation identifiers, PostMarX can read X bundle text and fetch a limited set of bundle files referenced by the page without credentials. The text is parsed only as data to find operation metadata and is never executed as extension code. All PostMarX executable code, WASM, and the included model are packaged with the extension. Optional downloads are reviewed model data and are never executed as extension code. X can change its page structure, so capture or import may require a PostMarX update. Use the Chrome Web Store listing's support channel if a workflow stops working.