Elasticsearch/OpenSearch MCP Server
Elasticsearch MCP connects multiple search clusters, with an explicit target name and access scoped to the supplied credentials.
- 315 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- By default the agent can delete indices and documents and send general API requests with the connected account’s permissions.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/elasticsearch-opensearch-mcp-serverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- By default the agent can delete indices and documents and send general API requests with the connected account’s permissions.
This is third-party code. Review the repository files before installing.
What it does
Elasticsearch/OpenSearch MCP Server exposes search and cluster-management tools to an agent. It reads documents, inspects index mappings and checks cluster health and statistics. A separate analyzer operation shows how text becomes search tokens. Writes are available by default, including index creation, document updates and deletion. For research, an administrator can hide write tools and issue read-only credentials.
Who it is for. Search-cluster administrators investigating query results and index health with an agent.
Good fit when
- Search misses a word, and the agent inspects how the analyzer splits it into tokens.
- Before changing a query, you inspect the index fields and settings.
- During an incident, an administrator collects node health and cluster statistics.
- Several test clusters share one connection, and you compare them by name.
Not a fit when
- The agent must be unable to delete data, but write tools remain enabled.
- The HTTP endpoint would be publicly accessible without authentication.
- You need a visual dashboard for direct human use without an agent.
Example request
Inspect the test index mapping and explain why this query misses the expected documents. Make no changes.MCP
- Transport
- stdio, sse, http
- Authentication
- key is optional
Environment variables
ELASTICSEARCH_API_KEYsecret- Elasticsearch API key scoped to the required operations.
ELASTICSEARCH_HOSTS- Comma-separated Elasticsearch hosts.
DISABLE_HIGH_RISK_OPERATIONS- Set true to hide write tools.
Limitations
Package selection depends on the cluster: the default targets Elasticsearch 8.x, while 7.x and 9.x have separate package names. OpenSearch uses its own package and username/password authentication. Certificate verification is disabled by default, so enable it explicitly for production connections. The HTTP server accepts unauthenticated access without `MCP_API_KEY`; restrict network reachability and configure a separate connection key.
How to disable. Remove the MCP connection and stop the server. Revoke the cluster credential if no other applications still use it.
FAQ
How do I remove write operations?
Set `DISABLE_HIGH_RISK_OPERATIONS=true` before startup. Creation, deletion and the general API request should disappear from the tool list.
Can I connect multiple clusters?
Yes. Define named configurations and pass `cluster` in requests. Verify the target before each operation.
Where does a request without a cluster name go?
To the cluster named in `DEFAULT_CLUSTER`, or the first configured cluster. Set the value explicitly to avoid sending a request to another environment.
Can I inspect a data stream?
Yes. The server reads data-stream information and searches its documents. Creating a stream requires a matching index template.
Does Claude Desktop require a network port?
No. A stdio connection launches a local process. HTTP and SSE provide separate network transports.
Discussion
No questions or reviews yet.
Related
MCP serversOfficial MCP server from the OpenSearch project exposes a cluster to agents: indexes, mappings, DSL search and shard state.
An MCP server gives agents search and answers over local documents using hybrid retrieval, reranking and section links.
Official MongoDB MCP server: queries and aggregations, index management, Atlas clusters and local deployments
An MCP server for searching and downloading academic papers from arXiv, PubMed, bioRxiv, Semantic Scholar and other open sources
MCP server that gives an agent access to MySQL: schema inspection and queries, read-only by default, writes behind flags
Redis's official MCP server: strings, hashes, streams, JSON and vector search from your agent
README in short
Deployment options include a Docker image and a Helm chart. Multi-cluster settings can live in a JSON file, with the agent selecting a cluster by name in each request.