Search

Search is the layer that works the moment you install the plugin. No provider, no API key, no index build to wait through. Open the command palette and run Smart Second Brain: Search notes.

Two retrieval paths

Search has a lexical path and, when an embedding model is configured, a semantic path. The search window runs one of them at a time. Only the agent can ask for both fused into a single ranking.

Lexical matches the words you typed. It is BM25-style scoring over a MiniSearch index, with fuzzy matching for typos and prefix matching so partial words still hit. It needs no model and no network.

Semantic matches meaning. Your query is embedded and compared against the embedded chunks of your notes, so a search for “why my mornings feel scattered” can surface a note that only ever says “context switching”. This path requires an embedding model. See What works without a model.

Results are normalized, boosted, and re-ranked before you see them. How search works documents the pipeline in full, including the constants.

Choosing a strategy

The search window always opens in lexical mode. Press Tab (or the semantic button in the tap bar on mobile) to run the current query semantically instead. That switch is a one-shot: the semantic results stay on screen, but as soon as you change the query text the next search runs lexically again. Switching requires a search embedding index. Without one the plugin shows a notice pointing at the setting.

Reach for lexical when you know the vocabulary: a project code name, a person, an error string. Reach for semantic when you know the idea but not the words your past self used.

The two modes are deliberately not fused in the search window. Pressing Tab means you have looked at the lexical results and they are not what you wanted, and mixing them back in re-injects the ordering you just rejected.

Strategy	What it matches	Needs an embedding model	Available to
`lexical`	Exact and fuzzy words	No	You and the agent
`semantic`	Meaning and concepts	Yes	You and the agent
`hybrid`	Both, fused	Yes	The agent only

hybrid exists only on the agent’s search_notes tool, and even there it is not the default: the agent starts lexical and is told to escalate, reaching for hybrid when a query mixes a specific term such as a name, tag, or filename with a fuzzy concept.

Filters

Searches can be narrowed before ranking rather than after:

Path: restrict to a folder or a single note.
Tags: restrict to notes carrying given tags.
Properties: restrict on frontmatter keys and values.

Filters apply at chunk level on the semantic path, so a filtered semantic search doesn’t quietly waste its result budget on chunks that were going to be discarded anyway.

Recency

Notes you have touched recently get a ranking boost. This is deliberate and mild: it breaks ties in favour of what you’re actually working on without letting a freshly-saved but irrelevant note outrank a strong match.

What search excludes

The agent’s own folder (Agents/ by default, holding memories, skills, and system prompts) is excluded from indexing, search, and the graph. It is plugin machinery that happens to be stored as notes, and it would otherwise dominate results about the plugin itself.

PDFs are indexed by their extracted text, not just their title — both for lexical search and, when an embedding model is configured, semantic search. Images are indexed by title only. There is no OCR or image-content extraction. On mobile, an oversized or scanned PDF falls back to title-only indexing, since above a certain size a PDF is predominantly scanned pages with no text to extract.

Chat files

Conversations are stored as .chat files in your vault, and they are indexed, so you can find a past conversation by searching for what was said in it. Because conversation history is a tree of checkpoints where each checkpoint re-contains every prior message, chat files are deduped by message id before indexing. Without that, a single conversation would be indexed many times over.

Chats are also transcludable: ![[My conversation.chat]] renders the transcript inside another note. See Embedding a chat in a note.

The `search_notes` tool

The agent reaches the same pipeline through its search_notes tool, with the same filters and one extra strategy, hybrid, which fuses the lexical and semantic paths. When you ask the agent a question about your notes, this is what runs. See Agents.