Leporello

Opera & classical music event schedule remote MCP.
Also available as a web app at leporello.app.

Supported Venues

Missing your favorite venue? PRs welcome! Read CONTRIBUTING.md for step-by-step instructions on adding a new venue scraper.

How it works

Venue-specific scrapers fetch schedule pages (HTML or JSON) from opera houses and concert halls, parse them with Cheerio, and store the events in a local SQLite database. A node-cron scheduler re-scrapes all venues daily at 03:00 UTC, replacing each venue's events with the latest data. The server exposes the data via a Model Context Protocol (MCP) endpoint, so any MCP-compatible AI assistant can query upcoming performances. A static Astro frontend reads the same database at build time and serves a filterable event listing at leporello.app.

MCP Tools

Privacy

The hosted instance at leporello.app logs aggregate MCP usage (tool name, arguments, response time, User-Agent, and a daily-rotating salted hash of the client IP) so I can see what's working and what isn't. No raw IPs, no personal data, no auth tokens.

Run locally

npm install
npm test          # run scraper tests

npm run scrape                       # one-off scrape all venues
#or
npm run scrape -- wiener-staatsoper  # scrape a single venue

npm run dev       # start server on http://localhost:3000
# or
npm run dev:fresh # scrape, then serve

Scrapes daily at 03:00 UTC. Run npm run scrape to populate the database on first use.

Docker

Two containers: web (HTTP server + MCP + static frontend) and scraper (fetches venue data into SQLite). They share a data volume. The web stack is docker-compose.yml; the scraper is a one-shot job in docker-compose.scrape.yml under its own compose project (leporello-scrape) so its exited container is never counted against the web stack's health. Schedule it with a host cron job.

# Start the web server
docker compose up -d

# Run all scrapers (one-off, container stops when done)
docker compose -f docker-compose.scrape.yml run --rm scraper

# Scrape a single venue
docker compose -f docker-compose.scrape.yml run --rm scraper node dist/scrape.js wiener-staatsoper

# Rebuild the Astro frontend (e.g. after a manual scrape)
docker compose exec web node -e "
import { execFile } from 'node:child_process';
import { promisify } from 'node:util';
await promisify(execFile)('npm', ['run', 'build', '--prefix', 'web'], { cwd: '/app', timeout: 60000 });
console.log('done');
"

# Tail logs
docker compose logs -f web

The web container rebuilds the Astro frontend on every start. To run scrapers on a schedule, add a host cron job that scrapes then restarts web:

0 3 * * * cd /home/phil/docker/lep && docker compose -f docker-compose.scrape.yml build && docker compose -f docker-compose.scrape.yml run --rm scraper && docker compose restart web