greptimedb-mcp-server

A Model Context Protocol (MCP) server for GreptimeDB — an open-source, cloud-native, unified observability database.

This server enables AI assistants to query and analyze GreptimeDB through SQL, TQL (PromQL-compatible), and RANGE queries. It includes security features like read-only enforcement and sensitive data masking, along with prompt templates for common analysis tasks.

Features

Resources

• list_resources - List all tables in the database as browsable resources
• read_resource - Read table data via greptime://<table>/data URIs

Tools

| Tool | Description | |------|-------------| | execute_sql | Execute SQL queries with format (csv/json/markdown) and limit options | | describe_table | Get table schema including column names, types, and constraints | | health_check | Check database connection status and server version | | execute_tql | Execute TQL (PromQL-compatible) queries for time-series analysis | | query_range | Execute time-window aggregation queries with RANGE/ALIGN syntax | | explain_query | Analyze SQL or TQL query execution plans | | list_pipelines | List all pipelines or get details of a specific pipeline | | create_pipeline | Create a new pipeline with YAML configuration | | dryrun_pipeline | Test a pipeline with sample data without writing to database | | delete_pipeline | Delete a specific version of a pipeline |

Prompts

MCP prompt system APIs:

• list_prompts - List available prompt templates
• get_prompt - Get a prompt template by name with argument substitution

Available prompt templates:

| Prompt | Description | |--------|-------------| | pipeline_creator | Generate GreptimeDB pipeline YAML configuration from log samples | | log_pipeline | Log analysis with full-text search and aggregation | | metrics_analysis | Comprehensive metrics analysis for monitoring data | | promql_analysis | PromQL-style queries using GreptimeDB TQL EVAL syntax | | iot_monitoring | IoT device monitoring with TAG/FIELD semantics and device aggregation | | trace_analysis | Distributed trace analysis for OpenTelemetry spans | | table_operation | Table diagnostics: schema, region health, storage analysis, and query optimization |

Using Prompts in Claude Desktop

In Claude Desktop, MCP prompts need to be added manually to your conversation:

1. Click the + button in the conversation input area
2. Select MCP Server
3. Choose Prompt/References
4. Select the prompt you want to use (e.g., pipeline_creator)
5. Fill in the required arguments

Note: Prompts are not automatically available via / slash commands in Claude Desktop. You must add them through the UI as described above.

LLM Instructions

Add this to your system prompt or custom instructions to help AI assistants use this MCP server effectively:

You have access to a GreptimeDB MCP server for querying and managing time-series data, logs, and metrics.

## Available Tools
- `execute_sql`: Run SQL queries (SELECT, SHOW, DESCRIBE only - read-only access)
- `execute_tql`: Run PromQL-compatible time-series queries
- `query_range`: Time-window aggregation with RANGE/ALIGN syntax
- `describe_table`: Get table schema information
- `health_check`: Check database connection status
- `explain_query`: Analyze query execution plans

### Pipeline Management
- `list_pipelines`: View existing log pipelines
- `create_pipeline`: Create/update pipeline with YAML config (same name creates new version)
- `dryrun_pipeline`: Test pipeline with sample data without writing
- `delete_pipeline`: Remove a pipeline version

**Note**: All HTTP API calls (pipeline tools) require authentication. The MCP server handles auth automatically using configured credentials. When providing curl examples to users, always include `-u <username>:<password>`.

## Available Prompts
Use these prompts for specialized tasks:
- `pipeline_creator`: Generate pipeline YAML from log samples - use when user provides log examples
- `log_pipeline`: Log analysis with full-text search
- `metrics_analysis`: Metrics monitoring and analysis
- `promql_analysis`: PromQL-style queries
- `iot_monitoring`: IoT device data analysis
- `trace_analysis`: Distributed tracing analysis
- `table_operation`: Table diagnostics and optimization

## Workflow Tips
1. For log pipeline creation: Get log sample → use `pipeline_creator` prompt → generate YAML → `create_pipeline` → `dryrun_pipeline` to verify
2. For data analysis: `describe_table` first → understand schema → `execute_sql` or `execute_tql`
3. For time-series: Prefer `query_range` for aggregations, `execute_tql` for PromQL patterns
4. Always check `health_check` if queries fail unexpectedly

Example: Creating a Pipeline

Ask Claude to help create a pipeline by providing your log sample:

Help me create a GreptimeDB pipeline to parse this nginx log:
127.0.0.1 - - [25/May/2024:20:16:37 +0000] "GET /index.html HTTP/1.1" 200 612 "-" "Mozilla/5.0..."

Claude will:

1. Analyze your log format
2. Generate a pipeline YAML configuration
3. Create the pipeline using create_pipeline tool
4. Test it with dryrun_pipeline tool

Security

All queries pass through a security gate that:

• Blocks DDL/DML operations: DROP, DELETE, TRUNCATE, UPDATE, INSERT, ALTER, CREATE, GRANT, REVOKE
• Blocks dynamic SQL execution: EXEC, EXECUTE, CALL
• Blocks data modification: REPLACE INTO
• Blocks file system access: LOAD, COPY, OUTFILE, LOAD_FILE, INTO DUMPFILE
• Blocks encoded content bypass attempts: hex encoding (0x...), UNHEX(), CHAR()
• Prevents multiple statement execution with dangerous operations
• Allows read-only operations: SELECT, SHOW, DESCRIBE, TQL, EXPLAIN, UNION, INFORMATION_SCHEMA

Data Masking

Sensitive data in query results is automatically masked to protect privacy:

Default masked column patterns:

• Authentication: password, passwd, pwd, secret, token, api_key, access_key, private_key, credential, auth
• Financial: credit_card, card_number, cvv, cvc, pin, bank_account, account_number, iban, swift
• Personal: ssn, social_security, id_card, passport

Configuration:

# Disable masking (default: true)
GREPTIMEDB_MASK_ENABLED=false

# Add custom patterns (comma-separated)
GREPTIMEDB_MASK_PATTERNS=phone,address,email

Masked values appear as ****** in all output formats (CSV, JSON, Markdown).

Installation

pip install greptimedb-mcp-server

# Upgrade to latest version
pip install -U greptimedb-mcp-server

After installation, run the server:

# Using the command
greptimedb-mcp-server --host localhost --port 4002 --database public

# Or as a Python module
python -m greptimedb_mcp_server.server

Configuration

Set the following environment variables:

GREPTIMEDB_HOST=localhost      # Database host
GREPTIMEDB_PORT=4002           # Optional: Database MySQL port (defaults to 4002)
GREPTIMEDB_HTTP_PORT=4000      # Optional: HTTP API port for pipeline management (defaults to 4000)
GREPTIMEDB_HTTP_PROTOCOL=http  # Optional: HTTP protocol (http or https, defaults to http)
GREPTIMEDB_USER=root
GREPTIMEDB_PASSWORD=
GREPTIMEDB_DATABASE=public
GREPTIMEDB_TIMEZONE=UTC
GREPTIMEDB_POOL_SIZE=5         # Optional: Connection pool size (defaults to 5)
GREPTIMEDB_MASK_ENABLED=true   # Optional: Enable data masking (defaults to true)
GREPTIMEDB_MASK_PATTERNS=      # Optional: Additional sensitive column patterns (comma-separated)

Or via command-line args:

• --host the database host, localhost by default,
• --port the database port, must be MySQL protocol port, 4002 by default,
• --http-port the HTTP API port for pipeline management, 4000 by default,
• --http-protocol the HTTP protocol for API calls (http or https), http by default,
• --user the database username, empty by default,
• --password the database password, empty by default,
• --database the database name, public by default,
• --timezone the session time zone, empty by default (using server default time zone),
• --pool-size the connection pool size, 5 by default,
• --mask-enabled enable data masking for sensitive columns, true by default,
• --mask-patterns additional sensitive column patterns (comma-separated), empty by default.

Usage

Tool Examples

execute_sql

Execute SQL queries with optional format and limit:

{
  "query": "SELECT * FROM metrics WHERE host = 'server1'",
  "format": "json",
  "limit": 100
}

Formats: csv (default), json, markdown

execute_tql

Execute PromQL-compatible time-series queries:

{
  "query": "rate(http_requests_total[5m])",
  "start": "2024-01-01T00:00:00Z",
  "end": "2024-01-01T01:00:00Z",
  "step": "1m",
  "lookback": "5m"
}

query_range

Execute time-window aggregation queries:

{
  "table": "metrics",
  "select": "ts, host, avg(cpu) RANGE '5m'",
  "align": "1m",
  "by": "host",
  "where": "region = 'us-east'"
}

describe_table

Get table schema information:

{
  "table": "metrics"
}

explain_query

Analyze query execution plan:

{
  "query": "SELECT * FROM metrics",
  "analyze": true
}

health_check

Check database connection (no parameters required).

Pipeline Management

list_pipelines

List all pipelines or filter by name:

{
  "name": "my_pipeline"
}

create_pipeline

Create a new pipeline with YAML configuration:

{
  "name": "nginx_logs",
  "pipeline": "version: 2\nprocessors:\n  - dissect:\n      fields:\n        - message\n      patterns:\n        - '%{ip} - - [%{timestamp}] \"%{method} %{path}\"'\n      ignore_missing: true\n  - date:\n      fields:\n        - timestamp\n      formats:\n        - '%d/%b/%Y:%H:%M:%S %z'\n\ntransform:\n  - fields:\n      - ip\n    type: string\n    index: inverted\n  - fields:\n      - timestamp\n    type: time\n    index: timestamp"
}

dryrun_pipeline

Test a pipeline with sample data (no actual write):

{
  "pipeline_name": "nginx_logs",
  "data": "{\"message\": \"127.0.0.1 - - [25/May/2024:20:16:37 +0000] \\\"GET /index.html\\\"\"}"
}

delete_pipeline

Delete a specific version of a pipeline:

{
  "name": "nginx_logs",
  "version": "2024-06-27 12:02:34.257312110"
}

Claude Desktop Integration

Configure the MCP server in Claude Desktop's configuration file:

MacOS

Location: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows

Location: %APPDATA%/Claude/claude_desktop_config.json

Option 1: Using pip installed command (recommended)

{
  "mcpServers": {
    "greptimedb": {
      "command": "greptimedb-mcp-server",
      "args": [
        "--host", "localhost",
        "--port", "4002",
        "--database", "public"
      ]
    }
  }
}

Option 2: Using uv with source directory

{
  "mcpServers": {
    "greptimedb": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/greptimedb-mcp-server",
        "run",
        "-m",
        "greptimedb_mcp_server.server"
      ],
      "env": {
        "GREPTIMEDB_HOST": "localhost",
        "GREPTIMEDB_PORT": "4002",
        "GREPTIMEDB_USER": "root",
        "GREPTIMEDB_PASSWORD": "",
        "GREPTIMEDB_DATABASE": "public",
        "GREPTIMEDB_TIMEZONE": "",
        "GREPTIMEDB_POOL_SIZE": "5",
        "GREPTIMEDB_HTTP_PORT": "4000",
        "GREPTIMEDB_MASK_ENABLED": "true",
        "GREPTIMEDB_MASK_PATTERNS": ""
      }
    }
  }
}

License

MIT License - see LICENSE.md file for details.

Contribute

Prerequisites

• Python with uv package manager
• GreptimeDB installation
• MCP server dependencies

Development

# Clone the repository
git clone https://github.com/GreptimeTeam/greptimedb-mcp-server.git
cd greptimedb-mcp-server

# Create virtual environment
uv venv
source venv/bin/activate  # or `venv\Scripts\activate` on Windows

# Install development dependencies
uv sync

# Run tests
pytest

Use MCP Inspector for debugging:

npx @modelcontextprotocol/inspector uv \
  --directory \
  /path/to/greptimedb-mcp-server \
  run \
  -m \
  greptimedb_mcp_server.server

Acknowledgement

This library's implementation was inspired by the following two repositories and incorporates their code, for which we express our gratitude：

• ktanaka101/mcp-server-duckdb
• designcomputer/mysql_mcp_server
• mikeskarl/mcp-prompt-templates

Thanks!