Media
4 items
Description
Compute recommendations from bookmarks and browsing history, with interfaces for push and analysis. GitHub(Issue/PR): https://github.com/Browser-bookmark-hub/Bookmark-Record-Recommend ────────────────────────── A powerful extension built around Bookmark Records, Smart S-Score Recommendations, and AI Push & Analyze: it computes bookmark records, active time, click rankings, related history, and recommendation signals locally, then lets you download or push the “raw data + computed results” to GitHub so AI can analyze them out-of-the-box with AGENTS.md / CLAUDE.md rules. It is also an ecosystem-related project for [Bookmark-Canvas](https://github.com/Browser-bookmark-hub/Bookmark-Canvas), with [exported JSON files] compatible with Bookmark Canvas import formats for temporary sections and notes. 1. Bookmark Records (Data Pool) - Active Time Tracking: Tracks your actual active reading time spent on bookmarked pages. - Usage Stats: Collects and reviews bookmark additions, click rankings, related browsing history, and time rankings. - Context Search: Quickly find target records, historical timelines, and contextual clues with one unified search box. 2. Bookmark Recommendations (S-Score Pool) - Formula-Driven Algorithm: Surfaces forgotten high-value bookmarks through S-Score parameters fueled by the Data Pool. - Review Manager: Fully control your recommendation flow with “Review Later”, “Skip”, and “Block”. 3. AI Push & Analyze (Local Computation Result Pool) - Local Precomputation: Computes records, rankings, relations, and recommendation signals locally to reduce the AI’s raw-data burden. - One-Click Push: Download or push the “raw data + computed results” to GitHub as traceable data packages. - Out-of-the-Box Analysis: Let AI read structured outputs directly and analyze them with AGENTS.md / CLAUDE.md rules. ────────────────────────── v0.4.9.2 - Optimized the “Add to Review” export-button UI for narrow layouts. - Fixed right-click export for orange-selected items in Related Records. - Launched a new ecosystem project [Bookmark-Canvas](https://github.com/Browser-bookmark-hub/Bookmark-Canvas) and refreshed the ecosystem showcase images. v0.4.9.1 - Fixed click ranking incorrectly showing “No bookmarks to analyze” during sidebar cold start. - Improved distinction between bookmark loading failures and users with no bookmarks, with progressive retries. - Fixed month, week, and day export scopes accidentally exporting the entire history library. - Fixed inaccurate export ranges for cross-month and cross-week selections. - Fixed export scopes drifting after navigating between dates. - Added timestamps to local and GitHub export filenames to prevent same-day overwrites. - Fixed GitHub configuration loading for manual exports initiated outside the Push & Analysis view. - Cleaned up and ignored .DS_Store files. v0.4.9 > [!NOTE] > - **Bookmark-first history refresh**: visits to saved bookmarks now use batched URL-level updates. Ordinary web visits no longer automatically calibrate the local history library, reducing scans when opening the sidebar and during daily tasks; manual calibration or pre-export sync can fill them in. v0.4.8 > [!NOTE] > - **Export ecosystem expansion**: Push & Analyze and manual exports can target GitHub, local files, or both. GitHub manual exports are organized under `manual-export/<category>/<date>/` and remain separate from normal sync and pull flows. Related Records now supports right-click context exports for any item, configurable before/after ranges, and preserved orange-selection semantics; Click Ranking adds single-item exports. > - **UX and documentation polish**: recommendation modes now include notes on backup/restore and S-score recalculation; favicon warm-up is capped at 150 unique domains; bookmark-event API audit and multilingual documentation groundwork were added.