Events Agent – Apify Actor

Apify Actor that finds events by location, date window, and keywords from Luma and Cerebral Valley, then outputs a normalized, deduplicated dataset. Use separate actors for Meetup or Google if needed.

Sources

• Luma (lu.ma) – city-specific event pages (SF, NYC, London, etc.)
• Cerebral Valley – cerebralvalley.ai/events, pre-filtered by city
• Extra URLs (optional) – any event or list page URLs you add in sources.extraUrls

How location filtering works

Layer	What it does
Luma city page	Builds a city-specific URL (e.g. lu.ma/sf) so only local events are listed
CV list pre-filter	Reads event-card text on the global CV page and drops cards that mention a conflicting city or "Online"
Post-crawl city filter	Rejects events whose location fields clearly belong to a different city
Post-crawl country filter	Rejects events from a different country (US/UK/CA normalized)
Online detection	Online-only events are excluded when a physical city is requested; hybrid events pass through

Input schema (summary)

Field	Type	Default
location	{ city?, region?, country? }	{} – city filters by city; country adds secondary filter
keywords	string[]	[] – annotates events with matchedKeywords
startDate	date string	today (America/Los_Angeles)
endDate	date string	start + 14 days
sources	{ luma, cerebralValley, extraUrls[] }	both on, no extras
maxEventsPerSource	1–50	25
headless	boolean	true
proxyConfiguration	Apify proxy object	optional
debug	boolean	false

Date formats accepted: YYYY-MM-DD, MM/DD/YYYY, DD.MM.YYYY. Start date must be on or before end date.

Run locally

1. Install and build

   npm install
   npm run build
   

2. Run with local storage

   # Linux/Mac
   export APIFY_LOCAL_STORAGE_DIR=./storage
   
   # Windows
   set APIFY_LOCAL_STORAGE_DIR=./storage
   
   npm start
   

   Put your input at ./storage/key_value_stores/default/INPUT.json or use the Apify CLI.

3. Example input (see .actor/input.json)

   {
     "location": { "city": "San Francisco", "region": "CA", "country": "USA" },
     "keywords": ["AI", "startup", "tech"],
     "sources": { "luma": true, "cerebralValley": true, "extraUrls": [] },
     "maxEventsPerSource": 25
   }
   

Run timeout and event count

Default run options: timeout 3000s (50 min), memory 2048 MB.

• Quick run: maxEventsPerSource: 5 → ~12 requests, finishes in a few minutes.
• Standard: maxEventsPerSource: 25 → ~52 requests, well within 3000s.
• Maximum: maxEventsPerSource: 50 → ~102 requests, fits within 3000s.

Output (normalized event)

Results are deduplicated (stable ID + fuzzy title/time/city match) and filtered to real events only.

Each dataset item includes:

• url – event page URL
• title – event title
• date – YYYY-MM-DD from start time
• time – HH:mm from start time
• company – primary company (organizer or first company involved)
• description – event description

Plus when available: startTime, endTime, venueName, address, city, region, country, lat, lon, isOnline, attendanceType, organizerName, companiesInvolved[], speakers[], tags, matchedKeywords, imageUrl, scrapedFromUrls, and more.

Scripts

• npm run build – compile TypeScript to dist/
• npm start – run dist/main.js
• npm run lint – ESLint on src/
• npm test – run Jest unit tests

Extending

Add src/sources/<name>.ts with enqueue* and parse*Page functions, then wire them in main.ts.