docmd-search provides a single command with flags for different modes of operation.

Usage

docmd-search [directory] [options]

If no directory is specified, the current working directory is used.

Commands

docmd-search ./docs

Indexes the directory and opens an interactive terminal search. If an index already exists, only changed files are re-indexed.

Index + web UI

docmd-search ./docs --ui

Indexes the directory and launches a browser-based search interface powered by docmd.

Settings

docmd-search --settings

Opens the settings TUI to change your embedding model or reconfigure global options.

Help

docmd-search --help

Version

docmd-search --version

Options

Flag Description
--ui Launch the web UI in the browser after indexing
--dev Verbose output — shows directory paths, model info, and debug details
--model <id> Override the embedding model for this run (does not modify config files)
--settings Open the settings TUI to configure model and options
--version, -v Print the version number
--help, -h Show help message

Examples

Basic indexing
Web UI
Model override
CI/CD
# Index the docs/ folder in current directory
docmd-search ./docs

# Index current directory
docmd-search .

# Index with no arguments (uses current dir)
docmd-search
# Index and launch browser UI
docmd-search ./docs --ui

# With verbose output
docmd-search ./docs --ui --dev
# Use a specific model for one run
docmd-search ./docs --model Xenova/bge-small-en-v1.5

# Use the highest quality model
docmd-search ./docs --model Xenova/all-mpnet-base-v2
# Index as part of a build pipeline
docmd-search ./docs
# The .docmd-search/ directory is the output artifact

Exit codes

Code Meaning
0 Success
1 Error — missing peer dependencies, invalid directory, or indexing failure

Environment

docmd-search respects the following environment:

  • Global config: ~/.docmd-search/config.json — model selection and wizard state
  • Project config: .docmd-search/config.json — per-project overrides
  • Model cache: ~/.docmd-search/ — downloaded ONNX model files
Peer dependencies

Embedding generation requires @huggingface/transformers and onnxruntime-node. If they are missing, the CLI prints installation instructions and exits with code 1.