Skip to content
Weaviate MCP Server

Weaviate MCP Server

Released
4 views
v1.38
BSD-3-Clause

Summary

Weaviate's MCP server, built into the database itself — hybrid search, schema inspection, tenant listing and batch upserts at /v1/mcp, scoped by RBAC.

Features

  • Built into the weaviate/weaviate binary — no separate server to deploy
  • Streamable HTTP at /v1/mcp on the same port as the REST API
  • Hybrid search combining vector similarity with BM25 keyword matching
  • Schema inspection and multi-tenant listing
  • Batch object upserts, gated behind a separate write-access flag
  • Disabled by default; both flags runtime-configurable without a restart
  • Bearer/API-key auth that respects Weaviate's standard RBAC

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Weaviate MCP Server" tools will be available.
{
  "mcpServers": {
    "weaviate": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-proxy",
        "http://localhost:8080/v1/mcp"
      ]
    }
  }
}

Description

Weaviate's MCP server is not a package you install; it is part of the database. From v1.38 the Model Context Protocol server is generally available inside the main weaviate/weaviate binary, served as a streamable HTTP endpoint at /v1/mcp on the same port as the REST API. For anyone already running Weaviate, that removes the usual sidecar: no separate process to deploy, version-match or keep credentialed.

Off by default, on without a restart

The server is disabled by default as a deliberate security posture. Set MCP_SERVER_ENABLED=true to turn it on, and MCP_SERVER_WRITE_ACCESS_ENABLED=true if you want the agent to be able to write. Both flags became runtime-configurable in v1.38, so you can grant or revoke agent access to a live cluster without a restart — useful when write access should be a temporary, deliberate act rather than a permanent deployment property.

Four tools, scoped by RBAC

Authentication uses Weaviate's existing bearer/API-key flow and every call respects the instance's standard RBAC permissions, so an agent inherits exactly the entitlements of the key it presents. The tools are deliberately few:

  • weaviate-collections-get-config — read the schema configuration of collections, so the agent can discover what exists before querying it
  • weaviate-tenants-list — enumerate tenants in multi-tenant collections
  • weaviate-query-hybrid — hybrid search combining vector similarity with BM25 keyword matching, which is the retrieval mode most production Weaviate setups actually use
  • weaviate-objects-upsert — batch insert and update objects, available only when write access is enabled
Clients

Claude Code adds it with claude mcp add-json using an HTTP transport and an Authorization header; Cursor and VS Code take an equivalent streamable-HTTP block in .cursor/mcp.json or .vscode/mcp.json. Claude Desktop needs an mcp-proxy bridge, since it expects STDIO rather than HTTP.

Status

Previewed in v1.37.1 and promoted to general availability in v1.38, released 25 June 2026 alongside the HFresh vector index. Weaviate is open source under BSD-3-Clause.

Related MCP Servers

New

Official AntV MCP server that turns a description of your data into a rendered chart — 26 visualization types from bar and line to sankey, mind-map, fishbone and geographic maps.

Featured

InfluxData's official MCP server for InfluxDB 3 — schema discovery, bounded SQL and InfluxQL queries, line-protocol writes and token administration across Core, Enterprise and Cloud.

Browse all MCP servers →