Connect any AI agent to Elastic Cloud with a single URL

オープン
  • 何か問題がありますか？
#397 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
25/100
issue の種類
機能追加
明瞭さ
説明が足りない
活発さ
活発
技術スタック
elasticsearch
領域
authentication, backend-api-design, cloud, search

調査の方向性

The issue names no repository files, tests, or implementation entry points. Start by decomposing the hosted MCP endpoint, OAuth 2.1 consent and token lifecycle, Serverless project access, and Cloud Console connection management. Done means a globally hosted endpoint supports the listed client, authorization, revocation, and discovery outcomes for Elastic Cloud Serverless.

索引モデルが issue の本文から書いたものです。

説明

Component: Elastic Cloud Serverless product-area:platform

What the feature is

mcp.elastic.co is Elastic's entry into AI clients and marketplaces. It's a single hosted MCP endpoint where any AI client can discover and invoke Elastic capabilities, authenticated once across all your Serverless projects.

Many developer platforms already offer hosted MCP endpoints, giving them a presence in AI clients and letting agents connect without shell access or per-deployment configuration. mcp.elastic.co gives Elastic the same: one URL, one OAuth consent, access to your Elastic Cloud Serverless stack from any MCP-compatible AI client.

Value proposition

Today, connecting an AI agent to Elastic requires per-deployment API key management and per-project configuration. There's no org-level entry point, no way to discover your projects from an AI client, and no standard way to authorize access without sharing a static, long-lived key. mcp.elastic.co changes that: one OAuth 2.1 consent flow covers all your Serverless projects and regions, with short-lived tokens and instant revocation.

Elastic's data and search capabilities are now reachable wherever AI agents operate, without shell access or a custom integration per project.

Expected outcome

  • mcp.elastic.co is live as a globally hosted MCP endpoint for Elastic Cloud Serverless
  • Users authenticate once via OAuth 2.1. Short-lived tokens replace static API keys, through a browser-based consent screen.
  • AI clients can search documents, run ES|QL queries, query logs and metrics, manage Serverless projects, and more
  • Works with any MCP-compatible AI client
  • Elastic is discoverable in major AI client marketplaces
  • Organization administrators can view and revoke all agent connections across the org, for security audits and employee offboarding
  • Initially targets Elastic Cloud Serverless. ECH support is planned as a follow-on.

Key user stories / use cases

  • Developer connecting an agent: A data engineer adds mcp.elastic.co to their AI client of choice, authorizes once through a browser consent screen, and the agent can search their Serverless Elasticsearch index, run ES|QL queries, and inspect their Kibana dashboards. No API key created, no per-project configuration required.
  • User managing connections: A user navigates to Connected Applications in the Cloud Console, reviews the agents they've authorized, and revokes access to one they no longer use. The agent's tokens are immediately invalidated across all their projects.
  • Marketplace discovery: A developer browsing an AI client's marketplace finds Elastic listed alongside other developer tools. They add mcp.elastic.co with one click. Elastic's full data and search capabilities are available in their workflow, no configuration needed.
主要言語
言語のデータがありません
スター
6
フォーク
1
30日以内にマージされた PR はありません

環境構築

このプロジェクトには開発コンテナ、Dockerfile、コントリビューションガイドがありません。まず README を読み、一般的な手順ははじめてのコントリビューションガイドを参照してください。

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

elastic/roadmap のほかの issue

elastic/roadmap の issue をすべて見る

似ている issue

Security の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。