I built a local-first bookmark manager that actually remembers what you saved

Description and use

Linkit turns plain URLs into searchable, summarized, and interconnected knowledge, improving recall and reducing clutter while keeping data local and private.

Problem

Users save links quickly but struggle to find them later because browser bookmarks only store URLs, lacking summaries, semantic search, and duplicate detection.

Outcome

Linkit transforms raw URLs into searchable, summarized, and interconnected knowledge, improving recall and reducing clutter while keeping data local and private.

Workflow

• 1. Install Linkit via Homebrew or download binaries.
• 2. Import bookmarks from browsers with one click.
• 3. When a link is saved, the app fetches the page, generates a summary and key takeaways using an OpenAI-compatible API or local Ollama model.
• 4. The content is indexed for vector similarity search.
• 5. Duplicate URLs and near-duplicates are detected and can be merged with one click.
• 6. Background scans check for 404s and silent changes.
• 7. Users can create AI-curated collections by describing a goal in plain language.
• 8. The knowledge graph visualizes relationships between bookmarks via tags, collections, and semantic similarity.

Evidence

A browser bookmark stores a URL and nothing else.

Source

Linkit works on the saved content instead.

Source

All bookmarks live in a local SQLite database on your machine.

Source

If you want multi-device sync, there's an optional Supabase backend with row-level security where every record is bound to your own auth.uid()

Source