VMP MCP Server

The Volcengine Managed Prometheus Service (VMP) is a new-generation cloud-native monitoring system that fully inherits and integrates with the open-source prometheus ecosystem. The VMP MCP Server provides functions such as prometheus workspace queries and metric queries, facilitating the natural-language-driven metrics query and analysis experience in scenarios like operational troubleshooting and data analysis.

Version	v0.1.0
Description	Natural-language-driven query and analysis of prometheus metric data
Category	Observability
Tags	Prometheus, Monitoring, Observability

Tools

This MCP Server product provides the following Tools (capabilities):

Tool 1: list_workspaces

Detailed Description

Query all workspace information in the specified region under the current account.

Input parameters required for debugging:

Input:

{
    "name": "list_workspaces",
    "description": "Query all VMP workspace instances information in the specified region.",
    "inputSchema": {
        "type": "object",
        "properties": {
            "region": {
                "default": "cn-beijing",
                "description": "Target region (e.g. cn-beijing, cn-shanghai, cn-guangzhou)",
                "type": "string"
            }
        }
    }
}

Output:

• Workspace list

Example Prompt most likely to trigger

Please list all VMP workspace instances in the cn-beijing region.

Tool 2: query_metrics

Detailed Description

Execute a specified PromQL instant query in the specified VMP workspace.

Input parameters required for debugging:

Input:

{
    "name": "query_metrics",
    "description": "Execute a specified PromQL query statement in the specified VMP workspace.",
    "inputSchema": {
        "type": "object",
        "properties": {
            "workspaceId": {
                "description": "The ID of the VMP workspace instance to query.",
                "type": "string"
            },
            "query": {
                "description": "PromQL query statement.",
                "type": "string"
            },
            "time": {
                "anyOf": [
                    {
                        "type": "string"
                    },
                    {
                        "type": "null"
                    }
                ],
                "default": null,
                "description": "Query time, in RFC3339 format or Unix timestamp, default is the current time.",
            },
            "region": {
                "default": "cn-beijing",
                "description": "Target region (e.g. cn-beijing, cn-shanghai, cn-guangzhou)",
                "type": "string"
            }
        },
        "required": [
            "workspaceId",
            "query"
        ]
    }
}

Output:

• Metric query results

Example Prompt most likely to trigger

Query the CPU usage at the current time in the VMP workspace instance b73766b5-2e63-4143-bcd1-8a1ba3a94746 in the cn-beijing region.

Tool 3: query_range_metrics

Detailed Description

Execute a specified PromQL query within a specified time range in the specified VMP workspace.

Input parameters required for debugging:

Input:

{
    "name": "query_range_metrics",
    "description": "Execute a specified PromQL query within a specified time range in the specified VMP workspace.",
    "inputSchema": {
        "type": "object",
        "properties": {
            "workspaceId": {
                "description": "The ID of the target VMP workspace instance.",
                "type": "string"
            },
            "query": {
                "description": "PromQL query statement.",
                "type": "string"
            },
            "start": {
                "description": "Query start time, in RFC3339 format or Unix timestamp.",
                "type": "string"
            },
            "end": {
                "description": "Query end time, in RFC3339 format or Unix timestamp.",
                "type": "string"
            },
            "step": {
                "anyOf": [
                    {
                        "type": "string"
                    },
                    {
                        "type": "null"
                    }
                ],
                "default": null,
                "description": "Query step, in duration format, optional. If not provided, it will be automatically calculated based on the query time range.",
            },
            "region": {
                "default": "cn-beijing",
                "description": "Target region (e.g. cn-beijing, cn-shanghai, cn-guangzhou)",
                "type": "string"
            }
        },
        "required": [
            "workspaceId",
            "query",
            "start",
            "end"
        ]
    }
}

Output:

• Metric query results

Example Prompt most likely to trigger

Query the top 3 pods in terms of CPU usage in the last hour in the VMP workspace instance b73766b5-2e63-4143-bcd1-8a1ba3a94746 in the cn-beijing region.

Tool 4: query_metric_names

Detailed Description

Query the list of metric names that match the specified filter conditions in the specified VMP workspace.

Input parameters required for debugging:

Input:

{
    "name": "query_metric_names",
    "description": "Query the list of metric names that match the specified filter conditions in the specified VMP workspace.",
    "inputSchema": {
        "type": "object",
        "properties": {
            "workspaceId": {
                "description": "The ID of the target VMP workspace instance.",
                "type": "string"
            },
            "match": {
                "anyOf": [
                    {
                        "type": "string"
                    },
                    {
                        "type": "null"
                    }
                ],
                "default": null,
                "description": "Series Selector, used to filter the matching metric range, following the standard Prometheus Vector Selector syntax, e.g.: {job=~\"kubelet\"}",
            },
            "region": {
                "default": "cn-beijing",
                "description": "Target region (e.g. cn-beijing, cn-shanghai, cn-guangzhou)",
                "type": "string"
            }
        },
        "required": [
            "workspaceId"
        ]
    }
}

Output:

• List of matching metric names

Example Prompt most likely to trigger

What are the CPU related metrics in the VMP workspace b73766b5-2e63-4143-bcd1-8a1ba3a94746 ?

Tool 5: query_metric_labels

Detailed Description

Query the list of all label names for a specified metric in the specified VMP workspace.

Input parameters required for debugging:

Input:

{
    "name": "query_metric_labels",
    "description": "Query the list of all label names for a specified metric in the specified VMP workspace.",
    "inputSchema": {
        "type": "object",
        "properties": {
            "workspaceId": {
                "description": "The ID of the target VMP workspace instance.",
                "type": "string"
            },
            "metricName": {
                "description": "The name of the metric to query.",
                "type": "string"
            },
            "region": {
                "default": "cn-beijing",
                "description": "Target region (e.g. cn-beijing, cn-shanghai, cn-guangzhou)",
                "type": "string"
            }
        },
        "required": [
            "workspaceId",
            "metricName"
        ]
    }
}

Output:

• List of metric label names

Example Prompt most likely to trigger

What are the labels of the container_cpu_usage_seconds_total metric in the VMP workspace b73766b5-2e63-4143-bcd1-8a1ba3a94746 ?

Compatible Platforms

Ark, Trae, Cursor, Claude Desktop, or other terminals that support MCP Server calls.

Service Activation Link (Entire Product)

https://console.volcengine.com/prometheus

Authentication Method

API Key (Signature Mechanism)

Installation and Deployment

System Dependencies

• Install Python 3.10 or higher.
• Install uv
  • MacOS/Linux

  curl -LsSf https://astral.sh/uv/install.sh | sh

  • Windows

  powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

环境变量

Environment Variable Name	Description	Default Value	Acquisition Method
VOLCENGINE_ACCESS_KEY	Volcengine Account ACCESS KEY	-	Volcengine Access Console
VOLCENGINE_SECRET_KEY	Volcengine Account SECRET KEY	-	Volcengine Access Console
VOLCENGINE_REGION	Volcengine Region	cn-beijing	-

Deployment

UV

{
  "mcpServers": {
    "mcp_server_vmp": {
      "command": "uv",
      "env": {
        "VOLCENGINE_ACCESS_KEY":"Your Volcengine access key",
        "VOLCENGINE_SECRET_KEY":"Your Volcengine secret key"
      },
      "args": [
        "--directory",
        "/<your local path to mcp-servers>/server/mcp_server_vmp",
        "run",
        "mcp-server-vmp"
      ]
    }
  }
}

UVX

{
  "mcpServers": {
    "mcp_server_vmp": {
      "command": "uvx",
      "env": {
        "VOLCENGINE_ACCESS_KEY":"Your Volcengine access key",
        "VOLCENGINE_SECRET_KEY":"Your Volcengine secret key"
      },
      "args": [
        "--from",
        "git+https://github.com/volcengine/mcp-server#subdirectory=server/mcp_server_vmp",
        "mcp-server-vmp"
      ]
    }
  }
}

License

volcengine/mcp-server is licensed under the MIT License