The Algorithms Python

What it is

The Algorithms Python is a public collection of algorithms and data structures, each one a Python file you can read. The README calls it “All algorithms implemented in Python” and says it is for education. Files sit in folders by topic, such as sorts, searches, graphs, dynamic_programming, ciphers and maths. Contributors write them, and CONTRIBUTING.md asks for type hints, docstrings and doctests, so most files come with examples you can run. The README warns that the implementations “may be less efficient than the implementations in the Python standard library.”

1. Find a topic in DIRECTORY.md or on the docs site
2. Open the file and read its docstring
3. Run its doctests on your own copy
4. Change the code and run them again

Who it suits

People who learn an algorithm best by reading working code: students, self-taught developers and anyone preparing for interviews. The README says the code is for learning, and GitHub’s topics for the repository include education, interview and practice. It also suits anyone who wants a first open source pull request, since CONTRIBUTING.md welcomes new implementations and better doctests.

Skip it if you need a library to import in production. The README says the implementations are for learning only and to “Use them at your discretion.”

Where to use it

You can read it anywhere you open a web page, and you can run a file wherever a new enough Python runs (pyproject.toml asks for 3.15 or later).

• GitHub: the README, the folders and each file with its docstring
• The docs site: https://thealgorithms.github.io/Python/, built with Sphinx from the repository
• A clone or a ZIP: open the folders in an editor and run a file’s doctests
• A dev container in Visual Studio Code: the repository ships its own setup
• Gitpod: the README carries a Ready-to-Code badge, though Mindket did not open it
• Chat: a Discord server linked from the README and a Gitter room linked from CONTRIBUTING.md

The repository has an AGENTS.md and three agent skill files, written for people who send or review pull requests and triage issues. Mindket’s notes show how to use Claude Code, Codex or Cursor as a tutor while you read the repository, and they mark which steps are Mindket’s.

How to set it up

Get it saves a zip of Mindket’s setup notes in your downloads folder. The code needs no install to read. To run a file you need Python and your own copy of the repository.

1. Read the README on the repository page. It is https://github.com/TheAlgorithms/Python . Or click Code, then Download ZIP, and unzip it (Mindket’s note: ordinary GitHub, not in the README).
2. Or clone it with the first block below.
3. Open the folder in an editor and pick a file from DIRECTORY.md.
4. Run its doctests with the second block, using the file’s path.

git clone https://github.com/TheAlgorithms/Python.git
cd Python

The next command is from CONTRIBUTING.md. Replace my_submission.py with the file’s path, such as sorts/bubble_sort_recursive.py.

python3 -m doctest -v my_submission.py

How to use it

Start from DIRECTORY.md, which lists every Python file under its folder, and open one. Read the docstring first. CONTRIBUTING.md says it should explain the code or link a source such as Wikipedia, and that doctests should cover both valid and erroneous input. So the doctests read as a list of inputs and the answer each must give. Run them. Then change the code and run them again to see what breaks.

Read the end of a file before you run it directly. maths/greatest_common_divisor.py asks you to type two integers, and web_programming/current_stock_price.py requests Yahoo Finance pages.

The grouping below is Mindket’s. The folder names are the repository’s.

With an agent, name the file and ask for an explanation or a check, not a rewrite.

Open sorts/bubble_sort_recursive.py and explain in plain words what each doctest checks. Do not change any file.

Example

You open sorts/bubble_sort_recursive.py from the Sorts list in DIRECTORY.md. It opens with “A pure Python implementation of the recursive bubble sort algorithm.” The function takes a type parameter, def bubble_sort_recursive[T: Comparable](arr: list[T]) -> list[T]:, and its docstring holds doctests such as:

>>> bubble_sort_recursive([5, 1, 4, 2, 8])
[1, 2, 4, 5, 8]
>>> bubble_sort_recursive([1, "two"])  # doctest: +ELLIPSIS
Traceback (most recent call last):
...
TypeError: ...

The second example is the file’s rule for bad input: mixing a number and a string must raise a TypeError. You run the doctests with the command above. In Mindket’s test run on 3 November 2026 they passed under Python 3.14 and 3.12, with this file and one other copied from the repository into an empty folder. Then you change the comparison in the function and run them again to see which examples fail.

For newcomers and experts

If you are new. An algorithm is a fixed method for one task, such as putting a list in order or finding an item in it. A data structure is a way to hold data, such as a linked list or a tree. A doctest is an example inside a function’s docstring: a line that starts with >>> is a call, and the line under it is what the call must return. Python’s doctest module runs those examples and reports any that differ. You need to be able to install Python and type a command in a terminal. Start with a folder whose name you know, such as sorts, and open the first file.

For experts. The repository tracks the newest CPython and its free-threaded builds. .python-version holds 3.15t, pyproject.toml requires 3.15 or later, and build.yml runs pytest with pytest-run-parallel and --ignore-gil-enabled. Its comment says some compiled dependencies do not yet ship the slot that keeps the GIL off, so importing them re-enables the GIL under 3.15t.

The build workflow’s pytest run skips four files that import cv2, the digital_image_processing folder, project_euler, docs/conf.py, scripts/validate_solutions.py and two web files. Pre-commit runs ruff, codespell, a filename check, rumdl for Markdown and uv-lock, among others. ty runs as an informational check that does not block a merge yet.

DIRECTORY.md is regenerated by scripts/build_directory_md.py at the end of a passing build. Project Euler answers are checked by scripts/validate_solutions.py, not by doctests, and that folder’s README says the ten slowest solutions show in the Actions logs. The scripts folder also holds shell scripts that close pull requests for missing tests, type hints or descriptive names, which its README ties to the Hacktoberfest rush in October.

Good to know

The code is for study. Read a file before you build on it.

• License: MIT. LICENSE.md begins “MIT License” and names “TheAlgorithms and contributors” as the copyright holders. project_euler/README.md says its problems come from projecteuler.net and “are licensed under CC BY-NC-SA 4.0”. Check that site before you reuse a problem statement
• Newer Python: in Mindket’s test run, maths/greatest_common_divisor.py stopped with a SyntaxError on Python 3.12 and passed on 3.14. The cause is an except line with no parentheses
• Stale setup files: .gitpod.yml and .devcontainer/Dockerfile still mention requirements.txt, which AGENTS.md says the project does not have. On 3 November 2026 GitHub’s raw address for it answered 404. The dev container configuration names an image and not that Dockerfile
• Skipped in CI: AGENTS.md says some directories are skipped on purpose (the --ignore entries in build.yml), usually because a heavy dependency lacks a wheel. Check that list to see whether a file is tested
• The web: a few files fetch pages, such as web_programming/current_stock_price.py. The dev container clones two zsh plugins from GitHub
• Sending a change: one algorithm file per pull request, and at least one box checked in its description, or the algorithms-keeper bot closes it. A vulnerability goes to the maintainers privately, as SECURITY.md says
• Outbound: reading loads GitHub, the README’s logo from raw.githubusercontent.com and its badges from img.shields.io. git clone and git pull talk to GitHub
• Removal: delete the folder, and the downloaded ZIP file if you kept it