Data Platform MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| DPMCP_MODE | No | Catalog adapter mode: 'demo' (default) or 'postgres'. | |
| DPMCP_LOKI_URL | No | Grafana Loki URL (e.g., https://loki.example.internal) when DPMCP_LOGS_MODE=loki. | |
| DPMCP_LOGS_MODE | No | Log adapter mode: 'demo' (default), 'opensearch', or 'loki'. | |
| DPMCP_TRANSPORT | No | Transport mode: 'stdio' or 'streamable-http' (default is stdio). | |
| DPMCP_LOKI_TOKEN | No | Grafana Loki API token when DPMCP_LOGS_MODE=loki. | |
| DPMCP_POSTGRES_DSN | No | PostgreSQL read-only connection string (e.g., postgresql://readonly_user:password@db:5432/analytics) when DPMCP_MODE=postgres. | |
| DPMCP_AIRFLOW_TOKEN | No | Airflow API token when DPMCP_OPERATIONS_MODE=airflow. | |
| DPMCP_OPERATIONS_MODE | No | Operations adapter mode: 'demo' (default) or 'airflow'. | |
| DPMCP_AIRFLOW_BASE_URL | No | Airflow instance base URL (e.g., https://airflow.example.internal) when DPMCP_OPERATIONS_MODE=airflow. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| healthA | Return MCP server health and version. |
| whoamiA | Return the authenticated principal, tenant, and effective RBAC scopes. |
| list_data_sourcesA | List data sources visible to the current tenant. |
| list_schemasB | List schemas in a configured and tenant-authorized data source. |
| list_tablesB | List tables or views in a tenant-authorized schema. |
| describe_tableC | Describe table columns and data types. |
| table_statisticsA | Return lightweight table statistics without scanning business rows. |
| get_table_metadataC | Return governed metadata for a tenant-authorized table or view. |
| get_table_lineageC | Return catalog-backed upstream lineage within the tenant source boundary. |
| list_etl_pipelinesA | List normalized ETL pipeline IDs published by the ETL metadata producer. |
| get_etl_pipelineC | Return the producer-owned normalized ETL metadata document. |
| get_etl_pipeline_stepsB | Return deterministic ETL step metadata for one pipeline. |
| get_etl_pipeline_dependenciesC | Return producer-classified ETL step/workflow dependencies. |
| get_etl_table_lineageC | Return producer-owned ETL lineage edges and capability boundaries. |
| search_etl_metadataC | Search ETL pipeline, step, table, and dependency names. |
| analyze_sql_lineageA | Parse read-only SQL and extract referenced input tables and CTEs. |
| explain_sqlB | Explain tenant-authorized read-only SQL after SQLGlot AST validation. |
| list_dagsA | List ETL orchestration DAGs known to the server. |
| get_dag_statusB | Return the latest known state of a DAG. |
| search_etl_logsC | Search the configured read-only ETL log backend. |
| search_runbooksC | Search operational runbooks and troubleshooting knowledge. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| incident_triage | Guide an MCP host through read-only DAG/log/runbook incident triage. |
| data_discovery | Guide metadata and lineage discovery before proposing analytics SQL. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| platform_capabilities | Active adapters, identity controls, telemetry, and safety mode. |
TDQS
Scored across 21 tools
Tools generally target distinct resources or aspects: data source/schema/table discovery, ETL pipeline metadata, lineage, SQL analysis, and observability. The closest overlap is among get_etl_table_lineage, get_table_lineage, and analyze_sql_lineage, but their producer/catalog/SQL scopes are described enough to avoid frequent misselection.
Most tools follow a consistent snake_case verb_noun pattern (list_*, get_*, search_*, describe_*). Minor deviations are health and whoami, which are bare commands, and table_statistics, which lacks a get_ prefix.
21 tools is slightly above the ideal 3-15 range, but the server spans several coherent subdomains: metadata discovery, ETL, lineage, SQL analysis, and operations. Each tool maps to a distinct operation, so the count does not feel padded.
For a governed, read-only metadata server the surface covers listing/discovery, table details, ETL lifecycle metadata, lineage, DAG status, logs, and runbooks. Minor gaps such as no dedicated getters for individual data sources/schemas and no downstream lineage are workaround-able but would improve full coverage.