About Teradata MCP Server
Teradata MCP Server gives MCP clients tools to query Teradata, inspect schemas, and analyze column values, with support for several authentication methods. It can also render query results as interactive charts.
Key features
- 01Run SQL queries and return tabular results
- 02List databases, tables, and column details
- 03Analyze missing, negative, and distinct values
- 04Calculate a column's mean and standard deviation
- 05Visualize query results with 19 chart types
- 06Support OAuth 2.1 and multiple database authentication methods
Use cases
- 01Query Teradata data from an MCP client
- 02Inspect table schemas and column types
- 03Review missing, negative, and category values
- 04Turn query results into interactive charts
MCP support
As an MCP server, it exposes SQL querying, schema browsing, data analysis, and chart visualization tools to clients. It supports stdio, SSE, and Streamable HTTP transports; OAuth 2.1 is available for HTTP transports.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMETeradata MCP Server
A Model Context Protocol (MCP) server for Teradata database with OAuth 2.1 authentication, multiple authentication mechanisms (TD2, LDAP, Kerberos), and interactive data visualization.
Features
- Multiple Auth Mechanisms — TD2 (default), LDAP, Kerberos, JWT via Teradata
LOGMECH - OAuth 2.1 with Keycloak integration, JWT validation, scope-based authorization
- Protected Resource Metadata (RFC 9728 compliant)
- Interactive Visualization — ECharts-based MCP App with 19 chart types
- Connection Resilience — automatic retry with exponential backoff
- Non-blocking I/O — all DB operations run via
asyncio.to_thread() - Per-tool QueryBand — audit trail for Teradata workload management
Tools
Query Tools
query— Execute SQL queries, return plain tabular resultsvisualize_query— Execute SQL and render interactive ECharts charts via MCP App
Schema Tools
list_db— List all databaseslist_tables— List tables/views in a databaseshow_tables_details— Show column names and types for a table
Analysis Tools
list_missing_values— Columns with NULL value countslist_negative_values— Columns with negative value countslist_distinct_values— Distinct category counts per columnstandard_deviation— Mean and standard deviation for a column
MCP App — Interactive Visualization
The visualize_query tool renders results as interactive charts in the MCP client.
| Category | Charts |
|---|---|
| Bar | Basic, Grouped, Stacked, Horizontal, Sorted, Waterfall, Rounded, Polar |
| Line | Basic, Smooth, Area, Stacked Area, Step |
| Pie | Pie, Doughnut, Rose / Nightingale |
| Scatter | Scatter, Bubble |
| Mixed | Bar + Line |
Quick Start
Installation
git clone https://github.com/arturborycki/mcp-teradata.git
cd mcp-teradata
uv sync
Run with TD2 (Standard Authentication)
uv run teradata-mcp "teradatasql://user:password@host/database"
Or via environment variable:
export DATABASE_URI="teradatasql://user:password@host/database"
uv run teradata-mcp
Configuration
Claude Desktop
Add to your claude_desktop_config.json:
TD2 (Username/Password)
{
"mcpServers": {
"teradata": {
"command": "uv",
"args": [
"--directory", "/path/to/mcp-teradata",
"run", "teradata-mcp"
],
"env": {
"DATABASE_URI": "teradatasql://user:password@host/database"
}
}
}
}
LDAP Authentication
{
"mcpServers": {
"teradata": {
"command": "uv",
"args": [
"--directory", "/path/to/mcp-teradata",
"run", "teradata-mcp"
],
"env": {
"DATABASE_URI": "teradatasql://@host/database",
"DB_LOGMECH": "LDAP",
"DB_LOGDATA": "authcid=ldap_user password=ldap_password"
}
}
}
}
The authcid format depends on the LDAP directory:
| Directory | Format |
|---|---|
| Active Directory (Simple Bind) | [email protected] |
| Active Directory (DIGEST-MD5) | authcid=DOMAIN\username |
| OpenLDAP / Sun DS | authcid=username |
Kerberos Authentication
{
"mcpServers": {
"teradata": {
"command": "uv",
"args": [
"--directory", "/path/to/mcp-teradata",
"run", "teradata-mcp"
],
"env": {
"DATABASE_URI": "teradatasql://@host/database",
"DB_LOGMECH": "KRB5"
}
}
}
}
OAuth-Enabled Configuration
{
"mcpServers": {
"teradata": {
"command": "uv",
"args": [
"--directory", "/path/to/mcp-teradata",
"run", "teradata-mcp"
],
"env": {
"DATABASE_URI": "teradatasql://user:password@host/database",
"OAUTH_ENABLED": "true",
"KEYCLOAK_URL": "https://your-keycloak.example.com",
"KEYCLOAK_REALM": "teradata-realm",
"KEYCLOAK_CLIENT_ID": "teradata-mcp",
"KEYCLOAK_CLIENT_SECRET": "your-secret",
"OAUTH_RESOURCE_SERVER_URL": "https://your-mcp-server.example.com"
}
}
}
}
Environment Variables
Database Connection
| Variable | Description | Default |
|---|---|---|
DATABASE_URI | Teradata connection URL (teradatasql://user:pass@host/db) | — |
DB_LOGMECH | Authentication mechanism: TD2, LDAP, KRB5, TDNEGO, JWT | TD2 |
DB_LOGDATA | LDAP/JWT credentials (e.g., authcid=user password=pass) | — |
DB_SSL_MODE | TLS mode: ALLOW, PREFER, REQUIRE, VERIFY-CA, VERIFY-FULL | — |
DB_ENCRYPT_DATA | Enable transport encryption | true |
Connection Resilience
| Variable | Description | Default |
|---|---|---|
DB_MAX_RETRIES | Max reconnection attempts | 3 |
DB_INITIAL_BACKOFF | Initial backoff delay (seconds) | 1.0 |
DB_MAX_BACKOFF | Max backoff delay (seconds) | 30.0 |
MCP Transport
| Variable | Description | Default |
|---|---|---|
MCP_TRANSPORT | Transport: stdio, sse, streamable-http | stdio |
MCP_HOST | Bind address for HTTP transports | localhost |
MCP_PORT | Port for HTTP transports | 8000 |
MCP_PATH | Path for streamable-http | /mcp/ |
OAuth 2.1
| Variable | Description | Default |
|---|---|---|
OAUTH_ENABLED | Enable OAuth authentication | false |
KEYCLOAK_URL | Keycloak server URL | — |
KEYCLOAK_REALM | Keycloak realm name | — |
KEYCLOAK_CLIENT_ID | OAuth client ID | — |
KEYCLOAK_CLIENT_SECRET | OAuth client secret | — |
OAUTH_RESOURCE_SERVER_URL | Resource server URL | — |
OAUTH_REQUIRED_SCOPES | Required scopes (comma-separated) | — |
OAUTH_VALIDATE_AUDIENCE | Validate token audience | true |
OAUTH_VALIDATE_SCOPES | Validate token scopes | true |
OAUTH_REQUIRE_HTTPS | Require HTTPS for OAuth URLs | true |
CORS_ALLOWED_ORIGINS | CORS allowed origins | * |
OAuth Scopes
| Scope | Description |
|---|---|
teradata:read | Read access to database resources |
teradata:write | Write access to database resources |
teradata:query | Execute SQL queries |
teradata:admin | Administrative access |
teradata:schema | Schema management operations |
Transport Compatibility
| Transport | OAuth | Discovery Endpoints | Use Case |
|---|---|---|---|
| stdio | N/A | N/A | Claude Desktop, CLI clients |
| SSE | Full | Available | Web applications |
| Streamable HTTP | Full | Available | API integrations |
Discovery endpoints (when OAuth enabled):
/.well-known/oauth-protected-resource— RFC 9728 metadata/.well-known/mcp-server-info— MCP capabilities/health— Health check
Docker Deployment
Development
docker compose up -d
With OAuth
docker compose -f docker-compose.oauth.yml up -d
Build
uv build
Troubleshooting
Database connection issues:
- Verify
DATABASE_URIformat:teradatasql://user:pass@host/database - Check network connectivity to Teradata server
- For LDAP: ensure
DB_LOGMECH=LDAPandDB_LOGDATAare set correctly - Connection issues are automatically retried (configurable via
DB_MAX_RETRIES)
LDAP authentication fails:
- Verify the Teradata server has LDAP configured in TDGSS
- Check
authcidformat matches your directory type - Escape special characters in passwords (
@→\@, spaces → use quotes)
Permission denied errors:
- Verify user has required OAuth scopes
- Check Keycloak role assignments
visualize_queryrequiresteradata:queryscope (not justteradata:read)
Debug logging:
export LOG_LEVEL=DEBUG
uv run teradata-mcp
License
MIT License. See LICENSE for details.
Acknowledgments
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.