diff --git a/README.md b/README.md
index c9f2c67..5a7dbdd 100644
--- a/README.md
+++ b/README.md
@@ -1,2 +1,151 @@
-# 025-weather
+# Weather Data (Day 25)
 
+Reading CSV weather data in Python, first with the built-in `csv` library and then
+with `pandas`. It is the day-25 assignment of Udemy's *100 Days of Code: The
+Complete Python Pro Bootcamp*. The lectures read the weather of one week from
+`weather_data.csv`, collect the temperatures as whole numbers, and then show how
+`pandas` does the same in one line: a table is a `DataFrame` and a column of it is
+a `Series`. They convert both to a dictionary and a list, compute the average and
+the maximum temperature, select rows by a condition (the row of Monday, the row of
+the hottest day, Monday's temperature in Fahrenheit), create a `DataFrame` from a
+dictionary, and export it to a CSV file.
+
+The `csv` part has no runtime dependencies; `pandas` is an optional extra. The code
+keeps the names of the lectures (`data`, `temperatures`, `temp_list`, `data_dict`,
+`average_temp`, `monday`, `to_dict`, `to_list`, `mean`, `max`, `to_csv`) so that you
+can compare it with your own solution.
+
+> **Status:** in progress. The project foundation (configuration, constants,
+> documentation settings and continuous integration) is in place; the readers and
+> the analysis follow in the next milestones, see `docs/project-plan.md`.
+
+## Requirements
+
+- Python 3.13 or newer.
+- `git`, to clone the repository.
+- Optional: [Doxygen](https://www.doxygen.nl/) to build the source documentation.
+- No runtime dependencies. `pandas` (the optional extra `pandas`) is needed for the
+  second half of the lectures and is installed by the `dev` extra, together with
+  the development tools (`pytest`, `ruff`, `mypy`).
+
+## Set up
+
+Clone the repository, then create a local virtual environment `.venv` in its
+root, upgrade `pip` and install the project with its development tools. The
+`.venv` folder is ignored by git.
+
+```bash
+git clone https://git.tirsystem.com/Tirsvad-Udemy-100-days-of-code/025-weather.git
+cd 025-weather
+```
+
+### Windows powershell
+
+```powershell
+python -m venv .venv
+.\.venv\Scripts\Activate.ps1
+python -m pip install --upgrade pip
+python -m pip install -e ".[dev]"
+```
+
+If PowerShell refuses to run the activation script, allow scripts for this
+window only with `Set-ExecutionPolicy -Scope Process -ExecutionPolicy RemoteSigned`
+and activate again.
+
+### Linux debian
+
+Debian 13 or newer ships Python 3.13. On older releases install Python 3.13
+separately.
+
+```bash
+sudo apt install python3 python3-venv
+python3 -m venv .venv
+source .venv/bin/activate
+python -m pip install --upgrade pip
+python -m pip install -e ".[dev]"
+```
+
+### MacOS
+
+```bash
+brew install python@3.13
+python3.13 -m venv .venv
+source .venv/bin/activate
+python -m pip install --upgrade pip
+python -m pip install -e ".[dev]"
+```
+
+To install only what the `csv` part needs, leave out the extras:
+`python -m pip install -e .`. To add `pandas` without the development tools use
+`python -m pip install -e ".[pandas]"`.
+
+## Run
+
+With the virtual environment active, the program that prints the results of the
+lectures is added in a later milestone (`python -m weather`); until then the
+package can be imported:
+
+```bash
+python -c "from weather import constants; print(constants.DATA_FILE)"
+```
+
+## Run the tests
+
+The tests write only into temporary folders.
+
+```bash
+pytest
+```
+
+The same checks that continuous integration runs:
+
+```bash
+python -m pytest
+python -m ruff check src tests
+python -m ruff format --check src tests
+python -m mypy
+```
+
+## Continuous integration
+
+`.gitea/workflows/ci.yml` runs on Gitea Actions on every push and on every pull
+request. It sets up Python 3.13, upgrades `pip`, installs `.[dev]`, then runs
+`pytest`, `ruff` (lint and format check) and `mypy`. It does not build the source
+documentation: run `doxygen Doxyfile` yourself, as the next section shows.
+
+The workflow lives in `.gitea/workflows` and not in `.github/workflows`, so
+GitHub does not run it and pushing to a GitHub mirror needs no workflow
+permission.
+
+## Build the source documentation
+
+The source is documented with Doxygen comments and the `Doxyfile` in the root.
+Install Doxygen (`winget install DimitriVanHeesch.Doxygen` on Windows,
+`sudo apt install doxygen` on Debian, `brew install doxygen` on macOS), then:
+
+```bash
+doxygen Doxyfile
+```
+
+The HTML is written to `build/doxygen/index.html`. A warning fails the build.
+
+## Project layout
+
+```text
+.
+├── .gitea/workflows/ci.yml    continuous integration (Gitea Actions)
+├── docs/                      business case, plan, milestones, reviews
+├── src/weather/               the package
+│   ├── __init__.py
+│   └── constants.py           every constant of the lectures
+├── tests/                     pytest tests (conftest.py runs every test in a
+│                              temporary folder)
+├── Doxyfile                   source documentation settings
+├── LICENSE
+├── pyproject.toml             project configuration
+└── README.md
+```
+
+## License
+
+GNU Affero General Public License v3.0 only. See [LICENSE](LICENSE).