Quick Start

Index a repository and run your first search in under a minute.

1. Index a repository

   cd path/to/repo
   unch index --root .
   

   Real output from indexing gorilla/mux:

   $ unch index --root .
   Loaded model       dim=768
   Indexed 278 symbols in 16 files
   
2. Run your first search

   unch search "create a new router"
   

   $ unch search "create a new router"
   1. mux.go:32  0.7747
   2. mux.go:314 0.8135
   
3. Add details when you want richer output

   unch search --details "get path variables from a request"
   

   $ unch search --details "get path variables from a request"
   1. mux.go:466  0.7991
      kind: function
      name: Vars
      signature: func Vars(r *http.Request) map[string]string
      docs: Vars returns the route variables for the current request, if any.
   

OpenRouter flow

1. Save the token once

   unch auth openrouter --token sk-or-...
   
2. Index with OpenRouter embeddings

   unch index --root . --provider openrouter --model openai/text-embedding-3-small
   
3. Search with the same provider and model

   unch search --provider openrouter --model openai/text-embedding-3-small "create a new router"
   

Search modes in practice

Auto

Best default for natural-language queries.

unch search "create a new router"

Semantic

Use when you want embedding-driven matches only.

unch search --mode semantic "parse query parameters"

Lexical

Use when you already know the exact identifier or string.

unch search --mode lexical "ParseQuery"

What happens on first run

The first local llama.cpp index may:

• download the default embedding model
• fetch local yzma runtime libraries
• create ./.semsearch/

Each provider and model pair keeps its own active index snapshot. Rebuilding openrouter/openai/text-embedding-3-small does not replace the active llama.cpp/embeddinggemma snapshot until the new run finishes successfully.

If you want to keep state somewhere other than ./.semsearch, use --state-dir.

Next steps

• Index command reference
• Search command reference
• Embedding models
• Remote indexing