Tooling Commands

Developer tooling: IDE integration, shell completions, and AI agent integration.

ide setup

archetect ide setup

Sets up IDE support for authoring Lua archetypes:

1. Installs Lua type annotations (archetect.lua and archetect_modules.lua) into the Archetect data directory: ~/.local/share/archetect/lua/annotations. These give lua-language-server (LuaLS) autocompletion and type checking for the archetect.* API.

2. Creates a .luarc.json in the current directory if it looks like a Lua archetype — i.e. it contains both an archetype manifest (archetype.yaml, archetype.yml, archetect.yaml, or archetect.yml) and an archetype.lua. The generated file sets the Lua runtime to 5.4 and points workspace.library at the installed annotations:

   {
     "runtime.version": "Lua 5.4",
     "workspace.library": [
       "/home/jane/.local/share/archetect/lua/annotations"
     ]
   }

   The write is idempotent: if an existing .luarc.json is byte-identical, nothing happens; if it is missing or differs (for example, the annotations directory moved), it is rewritten. A manifest without an archetype.lua skips the .luarc.json step with a notice.

Run it once per machine to install annotations, and again inside each archetype directory you author. archetect check reports whether the annotations are installed.

completions

archetect completions <shell>

Generates a shell completion script on stdout.

Argument	Values
<shell>	bash, elvish, fish, powershell, zsh

# bash
archetect completions bash > ~/.local/share/bash-completion/completions/archetect

# zsh (put on your fpath)
archetect completions zsh > "${fpath[1]}/_archetect"

# fish
archetect completions fish > ~/.config/fish/completions/archetect.fish

mcp

archetect mcp

Starts an MCP (Model Context Protocol) server on stdio, exposing Archetect to AI agents such as Claude Code. This is not meant to be run interactively — configure it in your agent's MCP server settings:

{
  "mcpServers": {
    "archetect": {
      "command": "archetect",
      "args": ["mcp"]
    }
  }
}

In MCP mode, the shell execution policy is forced to Forbidden — archetypes cannot execute commands, and there is no flag or environment variable to override this.