> ## Documentation Index
> Fetch the complete documentation index at: https://docs.momentra.org/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP server overview

> A hosted Model Context Protocol server exposing the Momentra events pipeline to ChatGPT, Claude, and any MCP client.

Momentra exposes its events pipeline as an [MCP](https://modelcontextprotocol.io)
server. Point it at the venues and organizations you care about, describe what
you're looking for as an editorial brief, and your assistant can search their
published calendars, discover events that match your brief, and turn the results
into plans.

## Server details

| | |
| - | - |
| **Server name** | `momentra-events` |
| **Endpoint** | `https://mcp.momentra.org/mcp` |
| **Protocol** | MCP over HTTPS (Streamable HTTP / SSE), JSON-RPC 2.0 |
| **Transport** | Stateless — a fresh transport per request (`GET`/`DELETE /mcp` → `405`) |
| **Auth** | OAuth 2.1 + PKCE with Client ID Metadata Documents (CIMD) |
| **Health** | `GET /health` → `200 {"ok":true}` |
| **Compatible hosts** | ChatGPT connectors, Claude connectors, any MCP-compatible client |

## The three layers

Momentra has three layers, from simplest to most powerful:

<Steps>
  <Step title="Index + search">
    Add organizations' calendars (`index_org`) and search/read their events
    directly (`search`, `search_business_events`, `fetch`).
  </Step>

  <Step title="Brief-driven matching">
    Write (or auto-learn) an editorial **brief** — your include/exclude policy —
    and let Momentra find the events and organizations that fit it
    (`matching_events`, `matching_orgs`), refining with accept/reject curation.
  </Step>

  <Step title="Seed from an aggregator">
    Bootstrap a brief automatically from a trusted aggregator site whose events
    you already trust (`seed_brief`), then sharpen it by reviewing borderline
    events.
  </Step>
</Steps>

<CardGroup cols={2}>
  <Card title="Connect it" icon="plug" href="/mcp/connecting">
    Add the connector in ChatGPT or Claude.
  </Card>

  <Card title="Tool reference" icon="wrench" href="/mcp/tools">
    All tools, with inputs, returns, auth, and cost.
  </Card>

  <Card title="Workflows" icon="route" href="/mcp/workflows">
    Step-by-step recipes for the common jobs.
  </Card>

  <Card title="Billing" icon="coins" href="/mcp/billing">
    Credits, the daily free cap, and billing responses.
  </Card>
</CardGroup>

<Note>
  Each result is returned as MCP tool-result content (a JSON text block).
  `search`/`fetch` follow OpenAI's connector result shape, and account-scoped
  tools also populate `structuredContent`.
</Note>

## Example prompts

Momentra works with any MCP-compatible client, so the same event data can power
a one-off prompt or a repeatable workflow. Once a client is connected, these run
as-is:

* **Weekly newsletter** — "Find family-friendly events in my area this weekend
  and draft a newsletter section with the top 10."
* **Destination guide** — "Build a fall events guide for the city, grouped by
  neighborhood and category."
* **Event feed** — "Generate a JSON feed of upcoming music events for the next 30
  days."
* **Calendar build** — "Assemble a month-long calendar of free outdoor events in
  my region."
* **Local recommendations** — "Suggest things to do near this hotel for guests
  arriving Friday."
* **Content pipeline** — "Every week, pull new events from these venues and
  format them for our CMS."

The agent discovers and indexes the relevant sources, then searches, matches, and
assembles the output. See [Workflows](/mcp/workflows) for how these compose from
the underlying tools.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.