Installation
Lila Framework v2.0.0 requires Python 3.10+ (Starlette 1.x and Uvicorn 0.54
need it). Optional extras: [redis],
[images], [markdown],
[metrics], [optimize]
and [desktop].
Build-Free Styling
Lila ships a zero-dependency, build-free styling system: no Node.js, Vite or npm step is required to get started.
- Tailwind Play CDN: Development templates load the Tailwind Play CDN script, so utility classes work immediately. Google Fonts are not loaded by Lila; the bundled layouts use the system font stack.
- Production: the Play CDN is a development convenience. For production the
README recommends shipping a precompiled CSS file and pointing
public('css/tailwind.css')at it. - Custom Component Layer: the bundled component classes (cards, inputs, buttons, layouts) support light and dark mode out of the box.
- Customization & Removal: edit your layout at
resources/html/lila/base.jinjaand change or remove the{{ asset('css/tailwind.css') | safe }}call.
️ Application Configuration
Purpose: Configure and start the application. main.py is the main entry point where the application is configured and initialized.
Zero-Disk I/O Configuration
Lila is built with maximum performance in mind. Reading from disk on every request is a known bottleneck. Lila caches environment variables and in-memory locales directly at CPython speeds with zero request overhead.
from app.config import DEBUG, HOST, PORT, WORKERS
from app.routes.web.index import router as web_router
from app.routes.api.index import router as api_router
from lila.core.app import App
# api_marker
# auth_marker
# admin_marker
# Initialize Lila application with optional batteries
app = App(
routes=[web_router, api_router],
debug=DEBUG, # bool, not a string
security=True, # Security headers & path traversal shield
rate_limit=300, # 300 req/min per IP (None to disable)
logging=False, # True to log requests in app/logs/YYYY-MM-DD/
cors=True, # Credentials are OFF by default
gzip=True,
openapi=True, # Mounts /openapi.json and Swagger UI at /docs
debug_html=DEBUG, # Mounts GET/DELETE /debug in debug mode
)
Run the production server with lila run -w 2; the CLI
reads HOST/PORT from
.env and warns when more than one worker is combined
with the in-memory cache.