OpenMetadata MCP Server

MCP server for OpenMetadata — 156 tools covering metadata management, data lineage, search (incl. semantic), data quality, and more. Read-only by default.

Authored to the @us-all MCP Standard — token-efficient by design.

Token-efficient by design — extractFields response projection, OM_TOOLS/OM_DISABLE category toggles, and a search-tools meta-tool let you keep LLM context costs low across the 155-tool surface.

When to use this vs OpenMetadata's embedded MCP

OpenMetadata 1.12+ ships with an embedded MCP server. They are complementary:

| | OM 1.12 embedded MCP | @us-all/openmetadata-mcp (this) | |--|----------------------|-----------------------------------| | Tool count | ~10 (search, glossary basics, lineage, DQ, RCA, semantic search) | 154 (full CRUD across all entity types) | | Auth | OAuth2 / PAT, inherits OM Authorization Engine (RBAC) | JWT bot token + OPENMETADATA_ALLOW_WRITE gate | | Deployment | Embedded in OM server (marketplace install) | Standalone npm / Docker / npx | | OM version | 1.12+ only | 1.x compatible | | Best for | RBAC-aware AI agents, SSO orgs, governance flows | Bulk CRUD, automation, sample-data inspection, older OM clusters |

Use the embedded MCP when you need RBAC-aware governance with SSO. Use this server for bulk metadata operations, full entity CRUD parity, and OM clusters older than 1.12.

Quick Start

# npx (no install)
npx @us-all/openmetadata-mcp

# or install globally
npm i -g @us-all/openmetadata-mcp
openmetadata-mcp

Claude Code

claude mcp add openmetadata \
  -e OPENMETADATA_HOST=http://your-host:8585 \
  -e OPENMETADATA_TOKEN=your-jwt-token \
  -- npx @us-all/openmetadata-mcp

Claude Desktop / Cursor

Add to your MCP settings JSON:

{
  "mcpServers": {
    "openmetadata": {
      "command": "npx",
      "args": ["@us-all/openmetadata-mcp"],
      "env": {
        "OPENMETADATA_HOST": "http://your-host:8585",
        "OPENMETADATA_TOKEN": "your-jwt-token"
      }
    }
  }
}

Docker

docker run --rm -i \
  -e OPENMETADATA_HOST=http://your-host:8585 \
  -e OPENMETADATA_TOKEN=your-jwt-token \
  ghcr.io/us-all/openmetadata-mcp-server

Environment Variables

| Variable | Required | Description | |---|---|---| | OPENMETADATA_HOST | Yes | OpenMetadata server URL (e.g. http://localhost:8585) | | OPENMETADATA_TOKEN | Yes | JWT or Bot token for authentication | | OPENMETADATA_ALLOW_WRITE | No | Set to true to enable create/update/delete operations (default: false) | | OM_TOOLS | No | Comma-separated allowlist of tool categories to load (e.g. core,governance,quality). When set, only these load — biggest token saver. Categories: search, core, discovery, governance, quality, services, admin, events. | | OM_DISABLE | No | Comma-separated denylist (e.g. events,admin). Ignored when OM_TOOLS is set. |

Token Efficiency

With 155 tools, naive setup loads ~24K tokens of tool schema into LLM context before any conversation begins. Three patterns mitigate this.

Measured impact (from tools/list JSON length, ~4 chars/token):

| Scenario | Tools loaded | Schema tokens | vs default | |----------|--------------|---------------|-----------| | default (all categories) | 155 | 24,000 | — | | typical (OM_TOOLS=search,core,governance,quality,discovery) | 120 | 19,500 | −19% | | narrow (OM_TOOLS=search,core) | 26 | 4,600 | −81% |

extractFields response projection adds another ~90% reduction on individual tool responses (e.g. get-table 8KB → 200B with extractFields: "name,columns.*.name,columns.*.dataType").

1. Category toggles (biggest win)

# Only data exploration tools
OM_TOOLS=search,core,discovery,quality

# Or exclude write/admin paths
OM_DISABLE=admin,events,services

Use search-tools (always enabled) to discover which tools exist regardless of what's loaded — call it first to find the right tool, then re-launch with broader categories if needed.

2. extractFields response projection

Available on 28 read tools (search, list/get for tables/dashboards/pipelines/charts/topics/mlmodels/containers/glossaries/glossary-terms). Comma-separated dotted paths with * wildcard:

// Without: 8KB JSON with 50+ fields per column
get-table { "id": "..." }

// With: ~200 bytes, just what you need
get-table { "id": "...", "extractFields": "name,description,columns.*.name,columns.*.dataType" }

3. search-tools meta-tool

Don't memorize 155 tool names. Discover by query:

search-tools { "query": "lineage" }
→ get-lineage, get-lineage-by-name, add-lineage, delete-lineage

Tools (155)

Meta (1)

search-tools — discover tools by natural-language query (always enabled)

Search (3)

search-metadata suggest-metadata semantic-search

Tables (6)

list-tables get-table get-table-by-name create-table update-table delete-table

Databases (6)

list-databases get-database get-database-by-name create-database update-database delete-database

Database Schemas (6)

list-schemas get-schema get-schema-by-name create-schema update-schema delete-schema

Lineage (4)

get-lineage get-lineage-by-name add-lineage delete-lineage

Services (16)

list-database-services get-database-service get-database-service-by-name create-database-service update-database-service delete-database-service list-dashboard-services get-dashboard-service list-messaging-services get-messaging-service list-pipeline-services get-pipeline-service list-ml-model-services get-ml-model-service list-storage-services get-storage-service

Glossaries (12)

list-glossaries get-glossary get-glossary-by-name create-glossary update-glossary delete-glossary list-glossary-terms get-glossary-term get-glossary-term-by-name create-glossary-term update-glossary-term delete-glossary-term

Dashboards (6)

list-dashboards get-dashboard get-dashboard-by-name create-dashboard update-dashboard delete-dashboard

Pipelines (6)

list-pipelines get-pipeline get-pipeline-by-name create-pipeline update-pipeline delete-pipeline

Topics (6)

list-topics get-topic get-topic-by-name create-topic update-topic delete-topic

Charts (6)

list-charts get-chart get-chart-by-name create-chart update-chart delete-chart

Containers (6)

list-containers get-container get-container-by-name create-container update-container delete-container

ML Models (6)

list-ml-models get-ml-model get-ml-model-by-name create-ml-model update-ml-model delete-ml-model

Classifications & Tags (10)

list-classifications get-classification create-classification delete-classification list-tags get-tag get-tag-by-name create-tag update-tag delete-tag

Domains & Data Products (12)

list-domains get-domain get-domain-by-name create-domain update-domain delete-domain list-data-products get-data-product get-data-product-by-name create-data-product update-data-product delete-data-product

Users & Teams (9)

list-users get-user get-user-by-name list-teams get-team get-team-by-name create-team update-team delete-team

Access Control (4)

list-roles get-role list-policies get-policy

Data Quality (7)

list-test-suites get-test-suite get-test-suite-by-name list-test-cases get-test-case get-test-case-by-name list-test-case-results

Stored Procedures (6)

list-stored-procedures get-stored-procedure get-stored-procedure-by-name create-stored-procedure update-stored-procedure delete-stored-procedure

Queries (5)

list-queries get-query create-query update-query delete-query

Events (3)

list-events get-event-subscription get-event-subscription-by-name

Bots (3)

list-bots get-bot get-bot-by-name

Sample Data (6, read-only)

get-table-sample-data get-table-sample-data-by-name get-topic-sample-data get-topic-sample-data-by-name get-container-sample-data get-container-sample-data-by-name

Getting a Token

1. Open your OpenMetadata UI
2. Go to Settings > Bots
3. Create a new bot or use an existing one (e.g. ingestion-bot)
4. Copy the JWT token

License

MIT