4 stable releases

Uses new Rust 2021

2.0.15 Dec 2, 2022
2.0.14 Nov 30, 2022
2.0.12 Jul 29, 2022
2.0.10 Jun 20, 2022
0.1.14 Aug 30, 2021

#22 in Cargo plugins

Download history 7/week @ 2022-10-07 7/week @ 2022-10-14 15/week @ 2022-10-21 14/week @ 2022-10-28 69/week @ 2022-11-04 2/week @ 2022-11-11 3/week @ 2022-11-18 59/week @ 2022-11-25 42/week @ 2022-12-02 4/week @ 2022-12-09 10/week @ 2022-12-16 7/week @ 2022-12-23 26/week @ 2022-12-30 2/week @ 2023-01-06 12/week @ 2023-01-13 13/week @ 2023-01-20

55 downloads per month

Custom license

963 lines

Cargo Commander

The simple way of running commands

Crates.io GitHub Sponsors GitHub last commit (branch) Build and test Website


Cargo Commander serves to fill the gap in the cargo commands capabilities, namely not being able to run commands in a similar fashion the way npm does with scripts. But while I was at it I decided to add some extra functionality to it.

New: In addition to running commands specified in either Commands.toml, Cargo.toml or package.json, functionality to execute scripts similar to how cargo-script does is being worked on. You can try it by either running a local script, cargo cmd script.rs, or running a remote script, cargo cmd https://url.to.script. This is currently in the early beta stages and functions by running rustc input -o output, then executing the output, so it's currently limited to using the standard library and the script has to be contained within that singular file. More features to come!

Getting started

Either create your commands under a [commands] or [package.metadata.commands] section in Cargo.toml, or create a new Commands.toml file. They all use the same syntax. Cargo commander also parses the scripts section inside package.json if it's found. Normally scripts inside package.json are only allowed to be strings, but Cargo Commander parses package.json by converting from json to toml, meaning you can add all the same options in json as you can in toml.

# Install cargo-commander
cargo install cargo-commander
# Run your command
cargo cmd COMMAND

# Output of 'cargo cmd --help'
cargo-commander 2.0.15
A powerful tool for managing project commands


    COMMAND              Name of the command to run
    URL                  Downloads a script, compiles then runs it
    FILE                 Compiles a file then runs it
    <ARGUMENTS>...       Arguments to the command

    -h, --help           Print help information
    -f, --file PATH      Custom path to command file to parse
    -p, --parallel       Forces all commands to run in parallel


A command can either be a string or a command object using the below fields to customize its behavior.

cmd = String or Array, where an array can either contain string commands or other command objects
parallel = true/false, only makes a difference if the command object contains an array, makes all commands run in parallel
shell = String, the syntax is simply "program arg arg arg"
env = Array, an array of strings in the format "VAR=SOMETHING"
args = Array, an array of strings in the format "ARG=Default value", if no default is given an empty string is used
working_dir = String, path to the directory to use as working directory either relative to the command file or the current directory


This can be either a string, a command object or an array of command objects.

If cmd is a multiline string the contents of the command is saved to a temporary file that gets safely deleted after the program finishes. The arguments are then used to replace content within the string, and the only argument sent to the shell is the path to the temporary file. We can use this behavior together with the shell option to create a file whose absolute path gets passed as an argument to whatever program you specify as a shell. See the examples for how this might look.

command = "echo Basic usage"
command = ["echo As an array"]
command = { cmd = "echo Hello" }
command = { cmd = ["echo Hello", "echo World"] }
command = { cmd = [{ cmd = "echo And hello again" }] }
command = { cmd = { cmd = "echo Hello again" } }


Boolean, defaults to false. If the cmd of the command object is an array, all sub commands will be run at the same time.

command = { cmd = ["echo first", "echo second", "echo third"], parallel = true }


String. The path where the command is supposed to execute in.

command = { cmd = "ls", working_dir = "src" }
command = { cmd = "ls", working_dir = "path/to/folder" }


Array of strings in the format args=["arg","argument=Default"]. If an argument is a string without a default value set it'll simply be replaced with an empty string.

command = { cmd = "echo $name", args = ["name=World"] }


Array of strings in the format env=["variable=Value"]. Sets environment variables in the command. This is similar to how args works, but the difference is that env changes environment variables. This option is generally speaking not super useful, you probably want to use load_dotenv instead.

# Unix
command = { cmd = "echo $HELLO", env = ["HELLO=World"] }
# Windows
command = { cmd = "echo %HELLO%", env = ["HELLO=World"] }


Boolean, defaults to false. Allows you to load environment variables from a .env file. The .env file should be located in the same folder as the file that contains the command being run. This option is unaffected by the working_dir option.

# Create a .env file with the contents "HELLO=World"
# Unix
command = { cmd = "echo $HELLO", load_dotenv = true }
# Windows
command = { cmd = "echo %HELLO%", load_dotenv = true }


Integer. Which status code counts as a successful run. Normally we don't check the status code of the command, but with this option we can tell the command to keep repeating until it reaches a specific exit code. If you set this to until=0 it would mean that you keep running the command until you reach a status 0 exit code. With until=404 it would keep running until you reach code 404. If you want to avoid infinite looping you should set max_repeat as well.

command = { cmd = "echo Hello", until = 0 }


Integer. Minimum number of times the command is meant to run. If you run this together with until you'll always be running the command at least this number of times.

command = { cmd = "echo Hello", repeat = 2 }


Integer or float. Amount of time to sleep before running the command. If you use this together with any of the repetition based options this delay will be added before every run of the command.

command = { cmd = "echo Hello", delay = 2 }
command = { cmd = "echo Hello", delay = 3.7 }


Integer. Sets the maximum number of times the command is allowed to retry. This is mostly useful when running together with until.

command = { cmd = "echo Hello", repeat = 5, max_repeat = 1 }
command = { cmd = "echo Hello", until = 0, max_repeat = 1000 }


Opening documentation

I have a tendency to create multiple mdbook books for documenting my projects. It's really neat, but it can be a bit of a bother to open them all one by one. So what I do is put the command to open each document under a docs section, then run the section rather than each individual page, using the -p flag to make the section run in parallel.

# Commands.toml
crate_one = { cmd = "mdbook serve --open --port 9001", working_dir = "crates/one/docs" }
crate_two = { cmd = "mdbook serve --open --port 9002", working_dir = "crates/two/docs" }
crate_three = { cmd = "mdbook serve --open --port 9003", working_dir = "crates/three/docs" }
crate_four = { cmd = "mdbook serve --open --port 9004", working_dir = "crates/four/docs" }

Now we can open all documents using a single command!

cargo cmd -p docs

Passing a custom argument

Let's say you want to get a running shell inside a Kubernetes pod where you don't know the pod name beforehand, probably because the pod was created by e.g. a deployment or a cronjob. There is a kubectl command you know of that can get you a running shell inside the pod, the problem is that the command is pretty long and annoying to write every time, and copy pasting the command from somewhere else every time gets repetitive really fast.

# Commands.toml
shell = { cmd = "kubectl exec --stdin --tty $pod -- /bin/bash", args = ["pod"] }

Now we can always get a shell to our pod by simple running the below simplified syntax. Now instead of having to both find the name of your pod and copy it into the longer kubectl command, you can now easily remember that you have a shell command that takes the argument pod.

cargo cmd shell pod=my-pod-123-654

Running a script

With a mix of the shell option and the behavior we've set for when a command is a multiline string we can achieve running scripts written directly in your command.

# Using python -c
hello_py_c = { cmd = "print('Hello')", shell = "python -c" }
# Using python and multiline string and an argument
hello_py = { cmd = """import os
""", args = ["name=World"], shell = "python" }

You can then run it as follows:

cargo cmd hello_py_c
# Or multiline
cargo cmd hello_py
# ... With argument
cargo cmd hello_py name=Commander

Keep retrying until command succeeds

Sometimes you run programs or write scripts that can fail. It's ok, it happens to everyone. Maybe it's a networked resource it's trying to reach, or maybe a file on your computer. No matter what the reason, the program will sometimes exit with a successful code 0, other times it exits with code 404 because the page it tried to reach wasn't found.

We can easily create a simple retry loop using until, combined with delay so that the program isn't ran too often, and max_repeat so that we don't try forever.

command = { cmd = "python script.py", until = 0, delay = 3, max_repeat = 1000 }

Running that command makes it keep retrying with a 3 seconds delay between retries. It will retry until it gets a 0 status returned, or a maximum of 1000 times.


Environment variables don't persist

I've tried to get this to work as intended but for now I've kind of given up on this since it appears to be anywhere between impossible and really, really annoying to get to work right. So each each command will have a "fresh" set of environment variables, if one command changes environment variables another command won't pick up on those changes, they are run in different shells. You can either use the env option, or you can run a script in every command that sets up environment variables, or you can use load_dotenv to load variables from a .env file. I consider these options to be sufficient, if you really want variables to persist across commands you'll have to make a pull request with your changes, or wait until I feel like delving deeper into the issue.


~775K SLoC