How to Use

This guide explains how to query monitoring data using an AI assistant after connecting to the WhaTap MCP server.

How it works

When you ask the AI assistant a question in natural language, it automatically handles the following steps.

1. Retrieve project list — Identify the project code (pcode).

2. Explore data catalog — Find what types of data are available.

3. Execute query — Retrieve the requested data.

Tip

Start with "Show me my project list" to confirm your project code. Using the code in follow-up questions will produce more accurate results.

Available tools

The MCP server provides 10 tools to the AI assistant. You do not need to call tools directly — just ask in natural language and the AI will automatically select the appropriate tool.

Project management tools

Tool	Function
whatap_list_projects	List all monitoring projects (pcode, name, platform)
whatap_project_info	Retrieve detailed information for a specific project
whatap_list_agents	List agents (servers/instances) within a project

Data exploration tools

Tool	Function
whatap_data_availability	Browse 900+ MXQL queries across 20+ categories, detect live data, list OpenMetrics metrics and saved PromQL queries
whatap_describe_query	Retrieve details for an MXQL path or OpenMetrics metric (parameters, output fields, metric type, recommended PromQL)
whatap_query_data	Execute MXQL, PromQL, or saved queries (results include unit conversion, thresholds, and analysis guidance)

Composite analysis tools

Tool	Function
whatap_apm_anomaly	APM anomaly detection — simultaneously runs 4 queries (TPS, response time, errors, Active TX) for statistical analysis
whatap_service_topology	Detect service connections and network bottlenecks (requires NPM agent)

Installation and PromQL tools

Tool	Function
whatap_install_agent	Generate agent installation commands for 29 platforms (auto-includes connection info, auto-detects platform)
whatap_create_promql	Create, validate, and save PromQL queries (reusable)

Supported platforms for whatap_install_agent

Category	Platforms
Infrastructure OS	Debian/Ubuntu, Amazon Linux, RHEL/CentOS, SUSE, FreeBSD, Windows, and 3 more (9 total)
APM	Java, Node.js, Python, PHP, .NET, Go
Database	PostgreSQL, Oracle, MySQL, Redis, MongoDB, DB2, and 6 more (12 total)
Container	Kubernetes (Helm)
Server apps	Kafka, NGINX, Apache, and 4 more (7 total)

Tool usage flow

Data query workflow (MXQL)

whatap_list_projects → Confirm project code
        ↓
whatap_data_availability(projectCode) → Check available MXQL paths
        ↓
whatap_describe_query(path) → Review parameters, fields, MXQL content
        ↓
whatap_query_data(projectCode, path) → Retrieve data

PromQL workflow (Kubernetes/OpenMetrics)

whatap_data_availability(projectCode) → Check OpenMetrics metric list
        ↓
whatap_create_promql(projectCode, name, query) → Validate and save PromQL
        ↓
whatap_query_data(projectCode, savedQuery="name") → Reuse saved query

Agent installation workflow

whatap_list_projects → Confirm projectCode + platform
        ↓
whatap_install_agent(projectCode) → Output connection info + installation commands
        ↓
whatap_list_agents(projectCode) → Verify agent registration

Tip

This flow is handled automatically by the AI. Just ask your question in natural language.

MXQL path reference

Commonly used paths you can query directly without calling whatap_data_availability first.

Domain	Path	Description
Server	v2/sys/server_base	CPU, memory, basic metrics
Server	v2/sys/server_disk	Disk usage
Server	v2/sys/server_network	Network I/O
APM	v2/app/tps_pcode	TPS (project-wide)
APM	v2/app/resp_time_pcode	Response time (project-wide)
APM	v2/app/tx_error_pcode	Error count (project-wide)
K8s	v2/container/kube_pod	Pod status
K8s	v2/container/kube_node	Node status
K8s	v2/container/kube_event	Cluster events
DB	v2/db/instance_active_session	Active sessions