Moss is a modern interpreted language with a simple C-style syntax.

Moss natively supports generation of output in many formats, such as Markdown, HTML and others. New formats and convertors can be also easily added.

Moss can also be compiled into bytecode for faster execution or non-source sharing.

This document was also generated by Moss and all code examples are run in "notebook" mode and output shown.

The example bellow showcases some nice Moss features, such as typed function arguments, ranges (start..end), calls to standard library functions (sum in space Math), list comprehension, documentation strings and outputting.

fun normalized_squares(start:Int, end:Int) { d""" Return list of squared numbers normalized to sum to 1. Numbers are in a range from `start` to `end`. """ squares = [x^2 : x = start..end] total = Math.sum(squares) return [x / Float(total) : x = squares] } 

normalized_squares(1, 6) 

[Output]:

[0.018182, 0.072727, 0.163636, 0.290909, 0.454545] 

As seen above in output for this function the return value was outputted. This is a big feature of Moss, where standalone expressions are outputted in specified format (in this case "text"). This is great for creating interactive notebooks and documents with runnable code that is recomputed on every generation. On top of that this output can be then easily converted to other formats (by specifying -f option followed by the desired format).

Moss has a built-in Python interoperability to take advantage of a wast library ecosystem which Python offers. This should also allow for easier implementation of Moss into existing projects which rely on existing Python scripts.

Working with python is very straight forward, Moss has PythonObject wrapper over Python objects. Here is an example of plotting a graph using matplotlib:

space { import python.* plt = module("matplotlib.pyplot") points = [18, 18, 25, 8, 18, 12, 25, 12, 1, 18, 0, 10, 20, 2, 18, 25, 25, 18, 33, 15] accum = Math.cumsum(points) races = [i : i = 1..points.length()+1] ~plt.plot(races, accum) ~plt.title("Max Verstappen - 2025 Season Cumulative Points") ~plt.xlabel("Race Number") ~plt.ylabel("Cumulative Points") ~plt.xticks(races) ~plt.grid(true) ~plt.savefig("docs/resources/ver_points.png") ~plt.close() } 

Moss also has built-in interoperability with C and any C-calling convention functions (see cffi module).

• Language reference documentation
• Wiki pages

Moss can be built and run on Linux, MacOS and Windows

• C++17 compatible compiler (gcc, clang),
• CMake,
• Python 3 (as Moss has a built-in interoperability).

Clone this git repository:

git clone --recursive https://github.com/mark-sed/moss-lang.git cd moss-lang

Moss can be built using CMake with specified targets or on Linux using pre-made install.sh script. This script requires sudo privileges to install into /usr/ and /bin/:

sudo bash instal.sh

With CMake it can be done with target moss and libms. If you wish to also install it onto correct path (to have command moss) you can use installation target.

On Linux:

cmake -S . -B build -DCMAKE_BUILD_TYPE=Release cmake --build build -j $(nproc) --target moss libms sudo cmake --build build -j $(nproc) --target installation

On MacOS:

On MacOS the base of the path is /usr/local/.

cmake -S . -B build -DCMAKE_BUILD_TYPE=Release cmake --build build --target moss libms installation

On Windows:

Open Developer Command Prompt and navigate into cloned moss-lang directory.

cmake -S . -B build cmake --build build --config Release --target moss libms installation

You can test this with:

# On Linux ./build/moss --version # On Windows build\\Release\\moss.exe --version

This file was generated on Nov 12 2025 by Moss 0.7.0 from readme.ms.