arc

arc (Automatic Remote Controller) is an infrastructure automation tool that uses Lua for scripting. It executes tasks on remote systems via SSH with a flexible API for managing configurations, files, and commands across multiple servers.

Installation

1. Install Rust

2. Install arc using Cargo:

cargo install arc-automation

This will compile and install the arc binary to the Cargo bin directory (usually ~/.cargo/bin/).

Please make sure you have the required dependencies installed:

Fedora

sudo dnf group install development-tools
sudo dnf install openssl-devel

Ubuntu / Debian

sudo apt install build-essential libssl-dev

Quick Start

Creating a New Project

Initialize a new arc project with type definitions for LSP support:

arc init /path/to/project

This command creates the project structure, type definitions for code completion and type checking, and a basic arc.lua file with example tasks.

Minimal Example

-- Define a target system
targets.systems["web-server"] = {
    address = "192.168.1.100",
    user = "root",
}

-- Define a simple task
tasks["hello"] = {
    handler = function(system)
        local result = system:run_command("echo 'Hello from ' $(hostname)")
        print(result.stdout)
    end
}

Run the task:

arc run -s web-server -t hello

See the examples directory for more complete usage examples.

Core Concepts

Targets

Targets define the systems where tasks will be executed. There are two types: individual systems and groups.

Systems

Systems can be either remote (accessed via SSH) or local (running on the arc host machine).

Remote Systems

Remote systems represent individual servers with SSH connection details.

targets.systems["frontend-server"] = {
    address = "192.168.1.100",
    user = "root",
    port = 22,  -- optional, defaults to 22
}

targets.systems["api-server"] = {
    address = "192.168.1.101",
    user = "deploy",
    port = 2222,
}

Local Systems

Local systems execute tasks on the machine where arc is running.

targets.systems["localhost"] = {
    type = "local"
}

Local systems use the same API as remote systems (same methods for run_command(), file(), directory(), etc.), but operations execute locally instead of over SSH. The address, port, and user properties return nil for local systems.

Groups

Groups organize multiple systems.

targets.groups["web-servers"] = {
    members = {"frontend-server", "api-server"}
}

targets.groups["prod"] = {
    members = {"prod-web-1", "prod-db-1"}
}

Tasks

Tasks define operations to execute on target systems. Tasks execute in definition order on each system.

tasks["install_nginx"] = {
    handler = function(system)
        local result = system:run_command("apt install nginx -y")
        if result.exit_code ~= 0 then
            error("Failed to install nginx: " .. result.stderr)
        end
    end,
    tags = {"nginx", "setup"},
}

tasks["configure_nginx"] = {
    requires = {"install_nginx"},
    handler = function(system)
        local config = system:file("/etc/nginx/nginx.conf")
        config.content = "..."
    end,
    tags = {"nginx"},
}

See Tasks API for all available fields.

CLI Reference

arc init

Initialize a new arc project with type definitions for LSP support, code completion and type checking.

arc init /path/to/project

arc run

Executes tasks defined in the arc.lua file.

Usage: arc run [OPTIONS] <--tag <TAG>|--all-tags> <--group <GROUP>|--system <SYSTEM>|--all-systems>

Options:
  -t, --tag <TAG>        Select tasks by tag
  -g, --group <GROUP>    Run tasks only on specific groups
  -s, --system <SYSTEM>  Run tasks only on specific systems
  -d, --dry-run          Print tasks that would be executed without running them
      --no-reqs          Skip resolution of requires and only run explicitly selected tasks
      --all-tags         Run all tasks
      --all-systems      Run on all systems
  -h, --help             Print help

Lua API Reference

arc uses a restricted LuaJIT environment. The following standard library modules are available:

• Modules (require)
• String Manipulation (string.format, string.match, string.gsub, etc.)
• Table Manipulation (table.insert, table.remove, table.sort, etc.)
• Mathematical Functions (math.floor, math.random, etc.)

Not available: io, os, debug, coroutine. Use the provided arc APIs (system:run_command(), system:file(), env.get(), etc.) instead.

The global print() function is an alias for log.info().

Tasks

Tasks are defined by assigning to the global tasks table. Tasks execute in definition order on each system.

Properties

• handler: Function that implements the task logic

  • Parameters: system - The system object to operate on
  • Returns: Optional result value accessible via tasks["name"].result

• tags (optional): Array of tags for filtering tasks. Tasks are automatically tagged with their name and source file path components (e.g., modules/web/nginx.lua adds tags: modules, web, nginx).

• targets (optional): Array of group or system names where this task should run. If omitted, runs on all systems.

• requires (optional): Array of tags this task requires. Tasks with matching tags are included when this task is selected. Resolved transitively.

• when (optional): Guard predicate that determines if the task should run

  • Returns: boolean - If false, task is skipped

• on_fail (optional): Behavior when this task fails

  • "continue" (default): Proceed to next task
  • "skip_system": Skip remaining tasks for this system
  • "abort": Halt execution entirely

• important (optional): If true, always runs regardless of tag filters, --no-reqs, and skip_system

State (read-only, available after execution)

• result: Return value from handler (nil if failed/skipped)
• state: "success", "failed", or "skipped"
• error: Error message if failed (nil otherwise)

Example:

tasks["check_nginx"] = {
    handler = function(system)
        return system:run_command("which nginx").exit_code == 0
    end
}

tasks["install_nginx"] = {
    requires = {"check_nginx"},
    when = function()
        return tasks["check_nginx"].result == false
    end,
    handler = function(system)
        local result = system:run_command("apt install nginx -y")
        if result.exit_code ~= 0 then
            error("Failed: " .. result.stderr)
        end
    end,
    on_fail = "skip_system",
}

Requires affect which tasks run, not when. Tasks always execute in definition order. If a task requires something defined later, the required task runs after the requiring task.

System Object

The system object represents a connection to a target system (remote or local) and is passed to task handlers.

Properties

• name: The name of the system as defined in targets.systems
• type: The type of system - "remote" or "local"
• address: The IP address of the system (nil for local systems)
• port: The SSH port of the system (nil for local systems)
• user: The SSH user used to connect to the system (nil for local systems)

Methods

• run_command(cmd): Execute a command on the system

  • Parameters: cmd (string) - The command to execute
  • Returns: A table with stdout, stderr, and exit_code

• file(path): Get a File object representing a file on the system

  • Parameters: path (string) - Path to the file
  • Returns: A File object

• directory(path): Get a Directory object representing a directory on the system

  • Parameters: path (string) - Path to the directory
  • Returns: A Directory object

Example:

tasks["check_service"] = {
    handler = function(system)
        log.info("Checking service on " .. system.name .. " at " .. system.address)
        local result = system:run_command("systemctl status nginx")
        return result.exit_code == 0
    end
}

File Object

The File object represents a file on a target system and provides access to file content, metadata, and operations.

Properties

• path: Path to the file (can be read and set; setting the path moves the file)
• file_name: The name of the file without the directory path (can be read and set)
• content: Content of the file as binary string (can be read and set)
• permissions: File permissions (can be read and set as numeric mode; returns nil if file doesn't exist)

Methods

• exists(): Check if file exists

  • Returns: boolean - true if file exists, false otherwise

• metadata(): Get file metadata

  • Returns: A table with file metadata (see Metadata Structure), or nil if file doesn't exist

• remove(): Remove the file

• directory(): Get the directory containing this file

  • Returns: A Directory object, or nil if at root path

Example:

tasks["configure_nginx"] = {
    handler = function(system)
        local config_file = system:file("/etc/nginx/sites-available/default")

        config_file.content = "server {\n    listen 80 default_server;\n    root /var/www/html;\n}"
        config_file.permissions = tonumber("644", 8)

        local metadata = config_file:metadata()

        if metadata and metadata.size then
            log.info("File size: " .. metadata.size .. " bytes")
        end
    end
}

tasks["manage_file"] = {
    handler = function(system)
        -- Create, move and then delete a file
        local file = system:file("/path/to/file.txt")
        file.content = "New content"                 -- Write to file
        file.permissions = tonumber("755", 8)        -- Set permissions
        file.path = "/new-path/to/renamed-file.txt"  -- Rename file
        file:remove()                                -- Delete file
    end
}

Directory Object

The Directory object represents a directory on a target system and provides access to directory operations and contents.

Properties

• path: Path to the directory (can be read and set; setting the path renames the directory)
• file_name: The name of the directory without the parent path (can be read and set)
• permissions: Directory permissions (can be read and set as numeric mode; returns nil if directory doesn't exist)

Methods

• create(): Create the directory (including any missing ancestor directories)
• remove(): Remove the directory
• exists(): Check if directory exists
  • Returns: boolean - true if directory exists, false otherwise
• metadata(): Get directory metadata
  • Returns: A table with directory metadata (see Metadata Structure), or nil if directory doesn't exist
• parent(): Get the parent directory
  • Returns: A Directory object representing the parent directory, or nil if at root path
• entries(): Get directory entries
  • Returns: Array of File and Directory objects representing the directory contents

Example:

tasks["setup_directory"] = {
    handler = function(system)
        -- Create directory structure
        local app_dir = system:directory("/var/www/myapp")
        app_dir:create()

        -- Set permissions
        app_dir.permissions = tonumber("755", 8)
    end
}

tasks["list_configs"] = {
    handler = function(system)
        -- Iterate through directory contents
        local dir = system:directory("/etc/nginx/sites-available")
        for _, entry in ipairs(dir:entries()) do
            -- Each entry is either a File or Directory object
            print(entry.path)
            print("Permissions: " .. entry.permissions)

            local metadata = entry:metadata()

            if metadata and metadata.type == "file" and metadata.size then
                print("File size: " .. metadata.size .. " bytes")
            elseif metadata.type == "directory" then
                print("Directory")
            end
        end
    end
}

tasks["manage_directory"] = {
    handler = function(system)
        -- Create, move and then delete a directory
        local dir = system:directory("/path/to/dir")
        dir:create()                                 -- Create directory
        dir.path = "/path/to/renamed-dir"            -- Rename directory
        dir:remove()                                 -- Delete directory
    end
}

Metadata Structure

The metadata() method on File or Directory objects returns a table with the following fields, or nil if the file/directory doesn't exist:

• path: Path to the file or directory
• size: Size in bytes (number, or nil if unavailable)
• permissions: Permission mode (number, or nil if unavailable)
• type: Type of the item ("file", "directory", or "unknown")
• uid: User ID of the owner (number, or nil; always nil on local systems)
• gid: Group ID of the owner (number, or nil; always nil on local systems)
• accessed: Last access time as a Unix timestamp (number, or nil if unavailable)
• modified: Last modification time as a Unix timestamp (number, or nil if unavailable)

Example:

tasks["check_metadata"] = {
    handler = function(system)
        local file = system:file("/etc/hostname")
        local metadata = file:metadata()

        if metadata then
            log.info("File size: " .. (metadata.size or "unknown"))
            log.info("File type: " .. metadata.type)
            log.info("File permissions: " .. (metadata.permissions or "unknown"))
            log.info("Last modified: " .. (metadata.modified or "unknown"))
        else
            log.info("File does not exist")
        end
    end
}

Environment Variables (env)

The env module provides access to environment variables. arc automatically loads variables from .env files in the project directory. Variables defined in the .env file take precedence over already defined ones.

Methods

• get(var_name): Get the value of an environment variable
  • Parameters: var_name (string) - Name of the environment variable
  • Returns: Value of the environment variable (string) or nil if not set

Example:

tasks["deploy_app"] = {
    handler = function(system)
        local app_version = env.get("APP_VERSION") or "latest"
        local deploy_path = env.get("DEPLOY_PATH") or "/var/www"

        system:run_command("docker pull myapp:" .. app_version)
        system:run_command("docker run -d -v " .. deploy_path .. ":/app myapp:" .. app_version)
    end
}

Host Module

The host module provides functions for interacting with the local system where arc is running. It has the same interface as the system object but operates on the local machine and it's working directory is the one arc was executed in.

Methods

• run_command(cmd): Execute a command on the local system

  • Parameters: cmd (string) - The command to execute
  • Returns: A table with stdout, stderr, and exit_code

• file(path): Get a File object representing a file on the local system

  • Parameters: path (string) - Path to the file
  • Returns: A File object

• directory(path): Get a Directory object representing a directory on the local system

  • Parameters: path (string) - Path to the directory
  • Returns: A Directory object

Example:

tasks["deploy_from_local"] = {
    handler = function(system)
        -- Read a local template
        local config_template = host:file("templates/nginx.conf").content

        -- Write to remote system
        local remote_config = system:file("/etc/nginx/nginx.conf")
        remote_config.content = config_template

        -- Restart service
        system:run_command("systemctl restart nginx")
    end
}

tasks["backup_to_local"] = {
    handler = function(system)
        -- Read from remote
        local remote_config = system:file("/etc/app/config.json").content

        -- Save locally
        local backup_dir = host:directory("backups/" .. system.name)
        backup_dir:create()

        local backup_file = host:file("backups/" .. system.name .. "/config.json")
        backup_file.content = remote_config
    end
}

Format Module

The format module provides utilities for working with JSON data.

Functions

• to_json(value): Convert a Lua value to JSON

  • Parameters: value (any) - Value to convert
  • Returns: JSON string

• to_json_pretty(value): Convert a Lua value to pretty-printed JSON

  • Parameters: value (any) - Value to convert
  • Returns: JSON string

• from_json(json_string): Parse a JSON string to a Lua value

  • Parameters: json_string (string) - JSON string to parse
  • Returns: Parsed Lua value

Example:

tasks["manage_json_config"] = {
    handler = function(system)
        -- Read a JSON configuration file
        local config_file = system:file("/etc/myapp/config.json")
        local config = format.from_json(config_file.content)

        -- Modify configuration
        config.debug = true
        config.log_level = "info"

        -- Write back to the file
        config_file.content = format.to_json_pretty(config)
    end
}

tasks["update_api_config"] = {
    handler = function(system)
        -- Get current config from an API
        local result = system:run_command("curl -s http://localhost:8080/api/config")
        local api_config = format.from_json(result.stdout)

        -- Update configuration
        api_config.settings.cache_ttl = 3600

        -- Send updated config back to API
        local json_config = format.to_json(api_config)
        system:run_command('curl -X POST -H "Content-Type: application/json" -d \'' .. json_config .. '\' http://localhost:8080/api/config')
    end
}

Template Module

The template module provides template rendering capabilities using the Tera template engine.

Methods

• render(template_content, context): Render a template with given context
  • Parameters:
    • template_content (string) - Template content
    • context (table) - Variables to use for template rendering
  • Returns: Rendered template as string

Example:

tasks["configure_web_server"] = {
    handler = function(system)
        -- Load a template from a local file
        local template_content = host:file("templates/nginx.conf.template").content

        -- Define context variables
        local context = {
            worker_processes = 4,
            worker_connections = 1024,
            server_name = system.name .. ".example.com",
            document_root = "/var/www/" .. system.name,
            environment = env.get("ENVIRONMENT") or "production"
        }

        -- Render and deploy configuration
        local config = template:render(template_content, context)
        system:file("/etc/nginx/nginx.conf").content = config

        -- Validate and reload
        local validation = system:run_command("nginx -t")

        if validation.exit_code == 0 then
            system:run_command("systemctl reload nginx")
        else
            error("Nginx configuration is invalid: " .. validation.stderr)
        end
    end
}

Logging Module

The log module provides logging functions at various severity levels.

Functions

• debug(value): Log a debug message

  • Parameters: value (any) - Value to log

• info(value): Log an info message

  • Parameters: value (any) - Value to log

• warn(value): Log a warning message

  • Parameters: value (any) - Value to log

• error(value): Log an error message

  • Parameters: value (any) - Value to log

Example:

tasks["provision_database"] = {
    handler = function(system)
        log.info("Provisioning database on " .. system.name)

        local result = system:run_command("systemctl status postgresql")

        log.debug("systemctl exit code: " .. result.exit_code)

        if result.exit_code ~= 0 then
            log.warn("PostgreSQL not running, attempting to install")

            local install_result = system:run_command("apt-get install -y postgresql")
            if install_result.exit_code ~= 0 then
                log.error("Failed to install PostgreSQL: " .. install_result.stderr)

                error(install_result.stderr)
            end
        end

        log.debug("PostgreSQL installed and running")
    end
}

LSP Support

arc provides Language Server Protocol (LSP) support for Lua code editing with autocomplete, type checking, and inline documentation.

Setup

1. Install the Lua Language Server for the editor being used.

2. Initialize an arc project to generate type definitions:

arc init /path/to/project

The init command creates .luarc.json and type definition files that enable the Lua Language Server to recognize arc's API types.

Contributing

Contributions are welcome! Please feel free to submit any issue or pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.