Indexing

Indexing generates vector embeddings for your bookmarks using Ollama, enabling semantic search. This is a local operation — no data leaves your machine.

Usage

bm index                          # index recent bookmarks (default: max 1 year old)
bm index --max-age 6m             # only bookmarks from last 6 months
bm index --max-age 0              # index everything regardless of age
bm index --reindex                # force re-index all bookmarks
bm index --model custom-model     # use a different embedding model
bm index -p "Default"             # limit to a specific profile

Age Filtering

By default, only bookmarks from the last year are indexed. This keeps indexing fast and avoids spending time on old bookmarks that may be less relevant.

--max-age accepts the same formats as bm fetch:

Format	Meaning
Nd	N days
Nw	N weeks
Nm	N months (30 days)
Ny	N years (365 days)
0	No limit

Prerequisites

Before indexing, you need:

1. Fetched content — run bm fetch first so there's page text to embed
2. Ollama running — ollama serve or brew services start ollama
3. An embedding model pulled — ollama pull qwen3-embedding:0.6b

How It Works

Each bookmark is split into chunks and sent to Ollama for embedding:

• Chunk 0: metadata (title + URL + folder path)
• Chunk 1+: page content in ~24,000 character slices

The resulting vectors are stored in the database alongside the bookmarks.

Choosing a Model

Default model: qwen3-embedding:0.6b

Any Ollama embedding model works. Override with --model or the BM_EMBED_MODEL environment variable.

Model	Size	Speed	Notes
qwen3-embedding:0.6b	Small	Fast	Good for most use cases
nomic-embed-text	Medium	Medium	Good general purpose
mxbai-embed-large	Large	Slower	Higher quality

Use the same model for indexing and searching

The embedding model used during bm index must match the one used during bm search -s. Vectors from different models are incompatible — cosine similarity between them produces meaningless results. If you switch models, re-index with bm index --reindex.