14.14 Build a TDengine Demo in One Click with Natural Language

TDengine EasyUse is a "one-click" automation orchestration skill for TDengine. Users only need to describe their business scenario in natural language (for example: smart monitoring of a wastewater treatment plant), and the complete workflow of data modeling → asset catalog construction → real-time analysis rule setup → visualization panel configuration is completed automatically.

This chapter describes the core value of the skill, environment preparation, deployment steps, and a hands-on example. After completing this chapter, you will be able to reproduce a complete industry demo independently on your own machine.

Section	Description
14.14.1 Core Value	Pain points EasyUse solves and target users
14.14.2 Environment and Prerequisites	Hardware and software requirements
14.14.3 Deploy TDengine	Deploy the base environment with All-in-One in one click
14.14.4 Create an API Key	Prepare credentials for MCP calls
14.14.5 Edit the Login File	Provide login information required by EasyUse
14.14.6 Install an AI Agent	Install an agent and choose an appropriate model
14.14.7 Deploy the EasyUse Skill Package	Clone the repository into the agent's Skills directory
14.14.8 Configure the MCP Server	Configure the agent to connect to the MCP server
14.14.9 Generate a Demo in One Click	Generate a complete demo with one sentence
14.14.10 Hands-On Example: Wastewater Treatment Plant	Details of the generated demo

Note The examples in this chapter assume a single-machine environment, with all service addresses pointing to http://localhost:6042. If you access from another machine, replace localhost with the hostname or IP address of the deployment node.

---

14.14.1 Core Value

Building a demo system for an industrial internet platform traditionally involves at least four manual steps:

1. Upfront definition on the business side: The business team must first systematically sort out the application scenario, identify key monitoring objects, define the core metrics system, and clarify the scope and goals of the project;

2. Translation and mapping into technical language: The technical team must translate business requirements into data models executable by the platform, accurately mapping business semantics onto technical structures;

3. Multi-step implementation at the platform level: System construction involves multiple critical steps — configuring asset structures, connecting data collection, setting alert rules, and building monitoring dashboards — making the workload highly concentrated;

4. Repeated cross-team communication: Business and technical teams typically need multiple rounds of requirement confirmation and iterative adjustments before arriving at a deliverable system.

Even an experienced team spends days or even weeks on this entire process.

TDengine EasyUse compresses all of this work into a single sentence of natural language: customers or pre-sales engineers only need to describe the business scenario (for example "smart monitoring of a wastewater treatment plant"), and the skill automatically completes modeling, the asset tree, visualization dashboards, and real-time analysis, shortening demo delivery from "days" to "tens of minutes". For internal reporting, solution reviews, and project approval, it is entirely sufficient.

Target Users

Role	Use Case
Customers / Trial users	Want to quickly see "what can this system do for me", skipping tedious configuration and going straight to business results
Pre-sales engineers	Need to quickly deliver a complete, demonstrable system from scattered customer Excel/CSV files and verbal requirements
Implementation consultants	Generate a demo structure in natural language before a real project starts, then fine-tune based on that structure

---

14.14.2 Environment and Prerequisites

Before starting deployment, prepare the following environment:

Category	Requirement
Operating system	Linux (recommended), macOS, or Windows + WSL
Permissions	root on Linux; administrator privileges on Windows (required for All-in-One deployment)
CPU	x64 or arm64 with AVX2 support (required by TDgpt)
Ports	6042, 6030, 6038, 6055, 6060

---

14.14.3 Deploy TDengine

All EasyUse capabilities are implemented by calling TDengine IDMP backend APIs through MCP, so a reachable TDengine environment must be available first.

We recommend deploying with TDengine All-in-One; see Section 14.13.

14.14.3.1 Deployment Steps

1. Visit the TDengine download center to obtain the All-in-One installation script;

2. Run the one-click deployment command. The script automatically downloads dependency packages and installs all TDengine components on your machine;

3. After deployment completes, open the following URL in your browser:

http://localhost:6042

4. Log in with the initial account and activate the product license as prompted (see Section 14.12 License Management).

Note If you do not need TDmodel, you can continue using the default deployment-single-node-no-tdmodel.yaml manifest. EasyUse itself does not depend on TDmodel.

---

14.14.4 Create an API Key

EasyUse calls TDengine IDMP over the MCP protocol, authenticating with a user-level API Key carried in the request header. To create one, see Section 14.8.3.

14.14.4.1 Steps

1. Log in to TDengine IDMP and click the avatar → account entry at the top in the upper-right corner;

2. Switch to the API Key tab and click Add API Key;

3. Enter a unique title (for example idmp-easyuse), then choose Never expires or specify an expiration date;

4. Click Create. A dialog pops up showing the full API Key (in the form api_xxxxxx);

5. Copy and store it immediately — the list page only shows a masked value afterwards.

Note An API Key inherits the role permissions and element access scope of its creator. If the key is later deleted or its owning user is deactivated, the API Key becomes invalid immediately. The request header format is fixed as Authorization: Bearer api_XXX — you do not need to manually prepend Bearer to the key itself.

---

14.14.5 Edit the Login Information File

During execution, EasyUse needs to read a plain-text login information file so that it can carry real credentials when writing to the database or creating models.

By default, open logint_info.txt in the idmp-easyuse directory and modify it according to your real user information:

TDengine IDMP:
host:     localhost
port:     6042
user:     <username>
password: <password>
api_key:  <API Key>

The fields correspond one-to-one with the TDengine IDMP login page; api_key uses the full value created in Section 14.14.4.

---

14.14.6 Install an AI Agent

TDengine EasyUse can run with any Skill-capable AI agent, such as Claude Code, Codex, OpenCode, WorkBuddy, etc. The agent is responsible for parsing natural language and dispatching skills to complete tasks.

This document uses OpenCode as an example:

14.14.6.1 Linux / macOS Installation

Run in a terminal:

curl -fsSL https://opencode.ai/install | bash
source ~/.bashrc         # or ~/.zshrc, depending on your shell

14.14.6.2 Windows Installation

On Windows, OpenCode officially recommends installing and using it inside WSL, to avoid path and shell compatibility issues in the native PowerShell environment. Follow the Linux procedure above within the WSL environment. If you need the native desktop version, download opencode-desktop-windows-x64.exe from the OpenCode GitHub Release page.

14.14.6.3 First Launch and Model Selection

1. Run in a terminal:

opencode

2. Enter the initialization command in the OpenCode interactive interface (optional; recommended on first use):

/init

3. Use /models to open the model list and select a large language model, for example:

DeepSeek V4 Flash

Note If the model list is empty, first configure the API Key of your model Provider in OpenCode (see the official OpenCode documentation). EasyUse is not bound to any specific model; any model supporting tool use works.

---

14.14.7 Deploy the EasyUse Skill Package

EasyUse consists of three skills working together. They need to be pulled from GitLab and placed into OpenCode's skill directory .opencode/.

14.14.7.1 Clone the Repository

git clone https://github.com/taosdata/agent-skills.git
cd agent-skills

14.14.7.2 Copy the Skills into OpenCode

Run in your OpenCode working directory (the directory containing a .opencode/ subdirectory):

cp -r skills/idmp-easyuse                  .opencode/
cp -r skills/idmp-sample-data-generator    .opencode/
cp -r skills/idmp-structure-exporter       .opencode/

The three skills divide the work as follows:

Skill	Purpose
idmp-easyuse	Master orchestrator: requirement parsing → data/asset modeling → real-time analysis configuration → visualization panel configuration
idmp-sample-data-generator	Generates the asset catalog and simulated time-series data matching the business semantics
idmp-structure-exporter	Exports/reuses an already-built asset catalog structure, for migration or replication

14.14.7.3 Verify Skill Loading

Back in the OpenCode session, run:

/skills

The list should include idmp-easyuse, idmp-sample-data-generator, and idmp-structure-exporter.

---

14.14.8 Configure the IDMP MCP Server

EasyUse communicates with IDMP through MCP. Edit the OpenCode configuration file ~/.config/opencode/opencode.json and add a server named idmp-mcp under the mcp section:

{
  "mcp": {
    "idmp-mcp": {
      "type": "http",
      "url": "http://localhost:6042/api/v1/mcp/stream",
      "headers": {
        "Authorization": "Bearer api_your_api_key_here"
      }
    }
  }
}

14.14.8.1 Field Description

Field	Value
Name	idmp-mcp (conventional name expected inside the EasyUse skill; do not change it)
URL	http://idmp-host:6042/api/v1/mcp/stream
Authentication	Authorization: Bearer with the API Key created in Section 14.14.4

Note If IDMP is deployed on another machine, replace localhost with the corresponding hostname or IP, and make sure port 6042 is reachable from the machine running OpenCode. Restart the OpenCode session after changing the configuration for it to take effect.

---

14.14.9 Generate a Demo in One Click

In the OpenCode session, invoke the EasyUse skill in natural language. The basic command format is:

/idmp-easyuse <business scenario description>.

Make sure not to specify any existing database! There is a risk that the specified database gets deleted during execution.

14.14.9.1 Complete Example

/idmp-easyuse create a wastewater treatment demo.

After execution, EasyUse proceeds in sequence:

1. Parses the business requirements → generates a Requirement Baseline;

2. Builds the asset model and asset tree;

3. Generates simulated time-series data and writes it;

4. Creates real-time analyses;

5. Creates visualization panels;

6. Outputs a demo generation report listing the paths of all created objects.

Once generation completes, you can view the asset tree, panels, and alerts directly at http://localhost:6042.

---

14.14.10 Hands-On Example: Wastewater Treatment Plant Smart Monitoring Demo

Taking the command /idmp-easyuse create a wastewater treatment demo… as an example, EasyUse created the following objects in one pass:

Make sure not to specify any existing database! There is a risk that the specified database gets deleted during execution.

14.14.10.1 Data Modeling (wastewater_treatment database)

The database wastewater_treatment is created automatically with 7 super tables designed:

Super Table	Collected Content
Water quality	pH, COD, ammonia nitrogen, total phosphorus, suspended solids, etc.
Flow	Inlet/outlet instantaneous flow, cumulative flow
Pumps	Running status, current, head
Valves	Opening degree, open/close status
Blowers	Rotational speed, air volume, vibration
Dosing	Chemical type, instantaneous/cumulative dosage
Energy	Active power, electricity consumption

14.14.10.2 Asset Tree (5 functional zones, 24 nodes)

Functional Zone	Typical Nodes
Pretreatment zone	Bar screen, grit chamber, primary sedimentation tank
Biochemical treatment zone	Anaerobic tank, anoxic tank, aerobic tank, secondary sedimentation tank
Advanced treatment zone	High-rate clarifier, sand filter, disinfection tank
Sludge treatment zone	Sludge thickening, dewatering machine, sludge pump
Auxiliary systems	Blower room, dosing room, power distribution, energy consumption

14.14.10.3 Visualization Panels (6)

For example:

• Aerobic tank dissolved oxygen and pH monitoring

• Blower running status monitoring

• Inlet vs outlet water quality comparison

• Dosing system dosage trends

• Sludge treatment energy consumption analysis

• Plant-wide key metrics overview

14.14.10.4 Alert Rules (8, across three severity levels)

Level	Example Rules
Critical	Outlet COD exceeding limit (> threshold for 5 consecutive minutes), sludge pump pressure too high
Major	Aerobic tank dissolved oxygen too low, abnormal blower vibration
Warning	Sudden inlet flow surge, dosing amount deviating from setpoint

All alerts come automatically with expressions, thresholds, and duration parameters, and can be edited again on the IDMP alerts page.

---

TDengine EasyUse transforms building an industrial data platform from "days of preparation, multi-person collaboration, repeated revisions" to "on the order of 30 minutes, started with a single sentence, straight into optimization". The change it brings is no longer just an efficiency gain but a restructuring of how projects get started — for the first time, industry experts' experience can quickly become systems, become results, become verifiable starting points.