Siftly

Siftly helps people with large Twitter/X bookmark collections find saved posts and organize them by topic. It runs on your own machine, with AI processing handled by the endpoint you choose.

• AI categorization: Sort bookmarks into default or custom categories using post text and visual context, with category descriptions guiding the assignments.
• Search across text and images: Find posts with natural language queries that cover tweet text, text detected in images, visual tags, and categories, with explanations for matches.
• Mindmap and browsing: Explore bookmarks in an interactive category graph, or use grid and list views with category, media, and date filters.
• Import and export: Import bookmarks without a browser extension, skip duplicates on repeat imports, and export CSV, JSON, or ZIP archives with media.

Siftly is open source under the MIT license and runs locally on macOS, Windows, and Linux, with Docker support. It stores its library in a local SQLite database and has no telemetry or tracking.

AI features support Anthropic, OpenAI, MiniMax, and local Anthropic-compatible endpoints. Cloud providers receive tweet text and image data for processing, so those features need an internet connection and provider authentication. On macOS, Siftly can use an existing Claude Code sign-in; Windows and Linux users can supply an API key.