* Fix: Stopped a single oversized media reference from aborting the whole scan.
* Update: Moved the unsupported plugins warning into the free version box as a compact note next to the Pro button.
* Fix: Trash recovery no longer requires a scan — recovered files are always available regardless of scan state.
* Add: Dashboard now shows a cleanup status message explaining when a fresh scan is needed.
* Update: Trash buttons are hidden while a scan is staged, preventing actions the server would refuse.
* Fix: Background images set in the Site Editor are now detected during scans.
* Fix: Hotfix — Results created before the scan-runs system are now properly adopted on update and reactivation, restoring cleanup and trash previews.
* Fix: MCP trashing no longer deletes files when the trash setting is skipped, and no longer reports unsaved progress.
* Update: Trash previews are now restricted to raster images.
* Update: Deleting results from an older scanner version now requires a fresh scan first.
* Add: MCP tools so an AI assistant can scan, review and clean up through AI Engine
* Add: Parser for background images stored in core block attributes
* Update: Reworked the references view into four meaningful columns with media titles, missing media markers, and fixed post titles and thumbnails
* Fix: References search returning no results because of an ambiguous column in the query
* Update: Replaced the browser broken image icon with a placeholder when a thumbnail cannot be loaded
* Add: Resumable, checkpointed scan runs with staged results that publish only after complete analysis.
* Add: Private quarantine storage, file identity validation, and idempotent cleanup operations.
* Fix: Graceful handling of timeouts, transient server failures, stalled cursors, and parser compatibility queries.
* Fix: Durable keyset pagination for large compatibility-parser datasets.
* Update: Full Cornerstone Parser coverage with URL references and improved data parsing.
* Update: Refactored URL detection to work with relative URLs.
* Add: Confirmation prompt before deleting the Cleaner database.
* Fix: Sped up file deletion by replacing directory scanning with a lazy emptiness check.
* Update: Made it clearer when viewing settings versus the dashboard.
* Add: In-results banner and tooltip clarifying that files marked as "not found in content" may still be used dynamically by builders, templates, or external sites.
* Fix: Memory overflow on large sites (1M+ media) by replacing in-memory array counting with SQL count queries.
* Fix: Added check for existing table before deleting attachment-related data.
* Fix: Meow Gallery parser now correctly applies WPLR filters on galleries and parses JSON HTML format.
* Add: Support for Meow Gallery's collections with WPLR integration.
* Add: Support for WP Download Manager plugin.
* Add: Filter "wpmc_affect_html_from_builder" for page builders that store their own content, allowing other parsers to still work alongside them.
* Update: Search now works across all reference fields.
* 🎵 Discuss with others about Media Cleaner on [the Discord](https://discord.gg/bHDGh38).
* 🌴 Keep us motivated with [a little review here](https://wordpress.org/support/plugin/media-cleaner/reviews/). Thank you!
* 🥰 If you want to help us, check our [Patreon](https://www.patreon.com/meowapps). Thank you!
* Update: Renamed internal origin references for clarity.
* Update: Path descriptions now use 'thumbnails' instead of 'files' for clarity.
* Fix: Image URLs in the metabox are now only added when actually present.
* Update: Improved UI/UX for better user experience. We hope you'll love it! 🥰
* Add: New "References Buffer" option to split and flush reference processing to reduce memory errors on large datasets.
* Fix: Improved base64 handling to better cope with malformed edge cases and prevent related issues.
* Add: More detailed timeout debugging for reference extraction to make diagnosing performance issues easier.
* Fix: Removed base64 content from HTML before URL parsing to prevent regex timeouts on complex pages.
* Update: Replaced cached references with a unique hash-based system and limited media ID lookups to multilingual URLs for faster and cleaner reference handling.
* Fix: Hotfix to resolve a security issue that could allow SSRF attacks through embedded iframes.