# Informa Events Directory MCP Server

MCP server for querying and retrieving structured event data from the Informa Connect event directory.

## Available Tools

Default (always available):

1. `get_informa_events`
Returns normalized events across all pages (API-first, HTML fallback).

2. `get_informa_events_page`
Returns a single directory page with normalized results.

3. `get_informa_events_by_date_range`
Returns normalized events constrained by `dateFrom` / `dateTo`.

4. `list_informa_event_facets`
Returns raw and simplified facet/filter metadata for the current query context.

5. `find_informa_event`
Finds one event by `id`, `slug`, `url`, or `name`.

6. `get_informa_event_detail`
Fetches one event page and extracts richer detail (JSON-LD event data, offers/pricing hints, contact hints, speaker extraction).

7. `get_informa_events_with_details`
Fetches directory events and bulk-enriches each with event-page details using configurable concurrency (including speaker extraction).

8. `get_informa_events_chunk`
Cursor-based chunked retrieval for large exports to avoid oversized single responses.

9. `get_informa_event_speakers`
Fetches one event page and returns speaker records only, with source selection (`merged`, `jsonld`, `redux`).

Feature-flagged (disabled by default):

10. `get_informa_discovery_suggestions` (feature-flagged)
Calls the discovery suggestions endpoint and returns raw JSON.

11. `search_informa_module_speakers` (feature-flagged)
Calls the discovery module speakers endpoint and returns raw JSON.

12. `search_informa_module_spex` (feature-flagged)
Calls the discovery module spex endpoint and returns raw JSON.

## Core Retrieval Behavior

- Primary data source: `POST https://informaconnect.com/api/v1/discovery/hub-search/pages/events`
- Fallback source: parse `window.__REDUX_STORE__` from HTML pages
- API payload optimization:
  - `includeFacets=false` (default) sends a disabled `facetsConfig` to reduce payload size
  - `includeFacets=true` enables facet payloads when needed
- Chunk export support:
  - `get_informa_events_chunk` returns `hasMore` + `nextCursor`
  - cursor can be passed back to continue where the previous chunk stopped
- Pagination stop conditions:
  - Empty page
  - Reached reported `totalPages`
  - `maxPages` safety cap
  - Two consecutive pages with no new unique events
- Retry/backoff for `429` and `5xx`
- Missing fields normalize to `null` or empty arrays

## Normalized Event Output

Each normalized event includes (if available):

- `eventName`
- `eventUrl`
- `description`
- `location.city`
- `location.venue`
- `location.country`
- `startDate`
- `endDate`
- `eventType`
- `subBrand`
- `industryOrCategory`
- `pricing` (currently usually `null` in discovery responses)
- `contactInfo` (currently usually `null` in discovery responses)

Plus additional enrichment fields (`eventId`, `sectors`, `topics`, `deliveryTypes`, etc.).

For detail tools, the event-detail payload also includes:

- `speakers[]` (merged/deduped from JSON-LD `performer` and Redux `speakers` when available)
- `speakerSources` (counts by source and total)

For speaker-only retrieval (`get_informa_event_speakers`), the response includes:

- `speakers[]`
- `speakerCount`
- `speakerSource` (requested mode)
- `speakerSources` (available totals by source)
- `eventMeta` (event name, dates, and location)

When `responseMode` is set to `"slim"`, each event is reduced to:

- `eventId`
- `eventName`
- `eventUrl`
- `startDate`
- `endDate`
- `location`
- `eventType`
- `subBrand`
- `sectors`
- `topics`
- `deliveryTypes`

When `responseMode` is set to `"summary"`, each event is reduced further to:

- `eventId`
- `eventName`
- `eventUrl`
- `startDate`
- `endDate`
- `city`
- `country`
- `venue`
- `eventType`
- `subBrand`
- `deliveryTypes`

## Install

```bash
npm install
npm run build
```

## MCP Config Example

Use [mcp-config.example.json](/Users/william.kitchin/Library/Mobile Documents/com~apple~CloudDocs/Codex Projects/Informa Events Directory MCP Server/mcp-config.example.json).

Example:

```json
{
  "mcpServers": {
    "informa-events-directory": {
      "command": "/absolute/path/to/informa-events-directory-mcp/build/index.js"
    }
  }
}
```

## Example Calls

### 1) Full crawl

```json
{
  "tool": "get_informa_events",
  "arguments": {
    "pageSize": 100,
    "maxPages": 50,
    "subBrandFilter": "Informa Connect",
    "eventTypeFilter": "EVENT",
    "requestedSortType": "DATE",
    "includeFacets": false,
    "responseMode": "slim"
  }
}
```

### 2) Single page fetch

```json
{
  "tool": "get_informa_events_page",
  "arguments": {
    "page": 2,
    "pageSize": 20
  }
}
```

### 3) Date-range query

```json
{
  "tool": "get_informa_events_by_date_range",
  "arguments": {
    "dateFrom": "2026-08-01",
    "dateTo": "2026-08-31",
    "pageSize": 100,
    "maxPages": 20,
    "countryFilter": "US",
    "responseMode": "slim"
  }
}
```

### 4) Find an event by slug

```json
{
  "tool": "find_informa_event",
  "arguments": {
    "identifier": "imn-btr-spring",
    "identifierType": "slug",
    "includeDetail": true
  }
}
```

### 5) Fetch event detail page

```json
{
  "tool": "get_informa_event_detail",
  "arguments": {
    "slug": "imn-btr-spring",
    "includeDirectoryRecord": true
  }
}
```

Detail response includes `detail.speakers` and `detail.speakerSources` when speaker metadata is present on the event page.

### 6) Bulk enrich many events with detail pages

```json
{
  "tool": "get_informa_events_with_details",
  "arguments": {
    "pageSize": 20,
    "maxPages": 5,
    "detailMaxEvents": 40,
    "detailConcurrency": 3,
    "stopOnDetailError": false
  }
}
```

### 7) Cursor-based chunk retrieval

```json
{
  "tool": "get_informa_events_chunk",
  "arguments": {
    "pageSize": 100,
    "chunkPages": 2,
    "requestedSortType": "DATE",
    "responseMode": "summary",
    "includeFacets": false
  }
}
```

Then continue with returned `nextCursor`:

```json
{
  "tool": "get_informa_events_chunk",
  "arguments": {
    "cursor": "<nextCursor-from-previous-response>",
    "chunkPages": 2
  }
}
```

### 8) Suggestions endpoint (feature-flagged)

```json
{
  "tool": "get_informa_discovery_suggestions",
  "arguments": {
    "searchInput": "biotech",
    "pageSize": 5
  }
}
```

### 9) Event speakers (single event)

```json
{
  "tool": "get_informa_event_speakers",
  "arguments": {
    "slug": "pharma-forum",
    "speakerSource": "merged"
  }
}
```

### 10) Speakers module endpoint (feature-flagged)

```json
{
  "tool": "search_informa_module_speakers",
  "arguments": {
    "speakers": [
      {
        "siteId": "example-site-id",
        "contentType": "SPEAKER",
        "displayType": "LIST"
      }
    ]
  }
}
```

### 11) Spex module endpoint (feature-flagged)

```json
{
  "tool": "search_informa_module_spex",
  "arguments": {
    "spexs": [{ "eventId": "c8b90e9b-548d-44a5-823f-f01ed5d08c13" }]
  }
}
```

## Useful Query Parameters

- `includeFacets` (`boolean`, default `false`): include discovery facets in API responses
- `responseMode` (`"full" | "slim" | "summary"`, default `"full"`): control event payload size
- `countryFilter` (`string | string[]`): alias merged into `locationFilter`
- `deliveryTypeFilter` (`string | string[]`): e.g. `"PHYSICAL"`, `"VIRTUAL"`, `"ON_DEMAND"`
- `sectorsFilter`, `topicsFilter`, `locationFilter`: facet-driven filtering inputs
- `cursor` + `chunkPages`: resume and control chunk-sized retrieval with `get_informa_events_chunk`
- `speakerSource` (`"merged" | "jsonld" | "redux"`): source selection for `get_informa_event_speakers`

## Speaker Retrieval Notes

- `get_informa_event_speakers` is the recommended tool when you only need speaker data.
- Speaker extraction works from event-page HTML, not directory listing API pages.
- Source priority for quality is typically JSON-LD `performer`, then Redux `speakers`.
- `speakerSource="merged"` returns deduped speakers from both sources.

## Response Metadata Notes

For multi-page tools (`get_informa_events`, `get_informa_events_by_date_range`, `get_informa_events_with_details`), response metadata includes:

- `totalResultsReported` and `apiReportedTotalCount`
- `totalPagesReported` and `apiReportedTotalPages`
- `isComplete` (whether retrieval completed naturally vs. capped by `maxPages`)

## Optional Environment Variables

```bash
INFORMA_ENTRY_URL="https://informaconnect.com/events/?subBrands%5B0%5D=Informa%20Connect&availableSubBrands%5B0%5D=Informa%20Connect&requestedSortType=FEATURED&page=1&count=20&searchInput=&type%5B0%5D=EVENT"
INFORMA_ORIGIN="https://informaconnect.com"
INFORMA_DEFAULT_PAGE_SIZE=20
INFORMA_DEFAULT_MAX_PAGES=200
INFORMA_REQUEST_TIMEOUT_MS=15000
INFORMA_MAX_RETRIES=4
INFORMA_ENABLE_SUGGESTIONS_TOOL=false
INFORMA_ENABLE_MODULE_TOOLS=false
```

Set `INFORMA_ENABLE_SUGGESTIONS_TOOL=true` and/or `INFORMA_ENABLE_MODULE_TOOLS=true` to expose the optional discovery tools.

## Limitations

- Discovery API is unofficial and may change.
- Some event metadata (pricing/contact) is not consistently present in discovery payloads.
- Facet metadata may be empty for some query contexts.
- HTML fallback relies on Redux state being embedded in the page.
- Anti-bot/CDN protections can affect reliability at higher request volume.
- `modules/speakers` request shape is not publicly documented and may return `422` unless exact speaker payload objects are provided.
- Speaker extraction quality depends on event-page metadata consistency; JSON-LD `performer` is usually the cleanest source.
- `search_informa_module_speakers` is included for experimentation, but event-page speaker extraction is currently the more reliable path.