# AGENTS.md Guidelines for AI coding agents working in this repository. ## Project overview A collection of Python CLI tools for working with OpenStreetMap data. Each tool is a standalone Python script using Click for the CLI interface and Requests for HTTP calls. ## Repository layout ``` pyproject.toml - Package metadata and build configuration src/ osm_geojson/ __init__.py - Top-level package marker pt/ __init__.py - Public transport subpackage core.py - Data fetching and processing functions cli.py - Click CLI commands tests/ fixtures/ - Saved OSM API responses for offline testing test_osm_pt_geojson.py - Test suite README.md - User-facing documentation AGENTS.md - This file ``` `osm_geojson` is the top-level namespace for all tools in this collection. Each tool lives in its own subpackage (e.g. `osm_geojson.pt`), with its CLI entry point registered in `pyproject.toml` under `[project.scripts]`. ## Code conventions - Python 3.11+ - CLI via [Click](https://click.palletsprojects.com/) - HTTP via [Requests](https://requests.readthedocs.io/) - Parse XML with lxml if needed; prefer the OSM JSON API where possible - Errors go to stderr; data output goes to stdout - GeoJSON output uses `ensure_ascii=False` - All modules, functions, and test functions must have docstrings ## OSM API Tools fetch data from the public OSM API: ``` https://www.openstreetmap.org/api/0.6/ ``` No authentication is required for read-only access. Include a descriptive `User-Agent` header in all requests. ## Type checking All code uses type hints. Run mypy in strict mode to check: ``` mypy --strict src/osm_geojson/ ``` ## Testing Run the test suite with: ``` pytest tests/ ``` Tests use the `responses` library to mock HTTP calls and never hit the live OSM API. Fixture data is stored in `tests/fixtures/` as saved API responses. ### Example relations used during development | ID | Description | |----------|------------------------------------------| | 15083963 | M11 Istanbul Metro (subway) | | 18892969 | Bus A1: Bristol Airport → Bus Station | ## Dependencies Use a virtual environment. Install the package and test dependencies with: ``` python3 -m venv .venv .venv/bin/pip install -e ".[dev]" ```