The open events layer for AI.

Verified, deduplicated, geo-structured event data for every covered city, updated daily — built to be read by LLMs and agents.

MCP endpoint

https://mcp.eventindex.ai

Paste into Claude, Cursor, or any MCP client.

Browse the live index → API docs →

Events live

deduplicated records

Cities covered

in 0 countries

Verified today

T-7 and T-1 cycles

Last pipeline run

not verified yet

pipeline warming up

60 min agorecords ingested per minutenow

Three ways in

same canonical record

One schema, three serializations. Whatever you read from, you get the same IDs, the same last_verified, the same canonical URL to cite.

01 · Directory

Crawlable pages

Ultra-light HTML with JSON-LD, and a markdown twin at the same URL plus .md. No JavaScript to execute.

eventindex.ai/it/milan/this-weekend
eventindex.ai/it/milan/this-weekend.md

Open the country index →

02 · MCP

Four tools, no install

Remote streamable HTTP. search_events, get_event, events_summary, resolve_city.

{
  "mcpServers": {
    "eventindex": {
      "url": "https://mcp.eventindex.ai"
    }
  }
}

MCP quickstart →

03 · REST API

Query the index

Events, nearby search, cities, city summary and a /changes feed. Key optional on the free tier.

curl "https://eventindex.ai/v1/events\
?city=milan&limit=3"

API reference →

Built to be crawled

robots.txt · verbatim

While most of the web blocks AI crawlers, we invite them in writing. Nothing is gated, nothing is cloaked, and the same content is served to bots and to people.

The crawler log is our own control panel: if GPTBot has not read a city hub in two weeks, that is the problem of the week.

AI crawlers welcome.

eventindex.ai/robots.txt

User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

User-agent: CCBot
Allow: /

User-agent: *
Allow: /
Crawl-delay: 1

Sitemap: https://eventindex.ai/sitemap.xml

Freshness is the moat

public last_verified

A stale event is worse than a missing one. Every record is re-checked on a schedule tied to its own start date, and the result is published on the page — relative and absolute, in UTC.

Daily sync

All active sources, every night

ingest, normalize, geocode, dedup, enrich, score, publish — 6 idempotent stages, at 03:23 UTC daily.

T-7 cycle

Seven days before doors

Price, time and venue re-read from the origin source; conflicts queue for review.

T-1 cycle

The day before

Cancellations and postponements land here. Honest states are published, never deleted.

On every page

The readout, in public

Relative age, absolute UTC timestamp, confidence and source count — on hubs and on events.

Access

4 tiers

Free

€0

1,000 requests/month · MCP unlimited, 60 req/min per IP

ATTRIBUTION REQUIRED

Starter

€49/mo

25,000 requests/month

Growth

€249/mo

200,000 requests/month · changes feed · webhooks

Scale

custom

From €1,000/mo · 99.9% SLA · no attribution

Full pricing →We never sell ranking. Access and scale only.