# Building with AI (/global/en/docs/Terminal/CodeDevelopment/Building-with-AI)

## Build on Newland PayExplorer Connect SDK with AI

We provide a standardized AI access tool suite centered on Skill packages, covering four core scenarios: transaction integration, API knowledge query, common problem troubleshooting, and access quality evaluation. Developers do not need to delve into complex documents; they only need to describe business requirements in natural language to obtain AI intelligent guidance, real-time problem solutions, and standardized access support. This greatly reduces the access threshold of PayExplorer Connect SDK, improves development efficiency, and shortens the integration cycle.

## Quick Start

### Install agent skills

#### npx (Recommended Method)

In the root directory of your project, directly run the following command to quickly install the required Skill package (the --yes parameter is used to automatically confirm the installation without manual interaction):

```bash
npx skills add https://git.newlandpayment.com/ai-projects/research-projects/skills/semi-integration-skill.git --yes
```

#### Newland Plugin market

Coming soon.

### Verify & Start

After installation, open AI IDE and enter the conversation interface, then input the guide command: `Help me integrate the functions of PayExplorer Connect`.

If the Skill package has been successfully activated, the AI assistant will provide accurate and actionable access guidance based on the professional knowledge in the Skill (instead of generalized answers); after verification, you can input more specific commands according to actual business needs to start AI-assisted development.

## Capabilities

### API Knowledge Query

Provide real-time retrieval service for full-scale API knowledge, covering interface parameter descriptions, business logic rules, error code interpretation, calling best practices and precautions. You can quickly obtain accurate information without manually browsing massive documents.

Sample Conversations:

- Please give an overall introduction to all interfaces of PayExplorer Connect
- What interfaces do I need to use to integrate consumption capabilities, and what are the required parameters for each interface
- What is the specific meaning and solution of the return code 4001 for the consumption query interface

### Transaction Integration

Based on the Skill package, it automatically generates standardized payment/transaction access code, covering full-scenario businesses such as consumption, cancellation, refund, transaction query, and settlement. The code can be directly reused or fine-tuned, greatly reducing manual coding workload and development errors.

Sample Conversations:

- I want to integrate bank card consumption function and generate the corresponding access code
- Help me connect the settlement function, including the complete code for settlement request and result callback
- Generate the interface call example and exception handling code for refund transaction

### Common Problem Troubleshooting

For common errors and abnormal scenarios during SDK integration, quickly query the cause, troubleshooting steps and solutions through Skill. You can efficiently solve development bottlenecks without repeatedly consulting technical support.

Sample Conversations:

- What is the reason for 'no target selected' returned during consumption, and how to solve it
- Why did the Abort request fail to cancel the transaction, and what is the troubleshooting idea
- The interface call prompts timeout, how to optimize the connection configuration

### Access Quality Evaluation

After the integration and development is completed, you can initiate a comprehensive access quality check through the AI assistant, covering four dimensions: interface usage standardization, correctness of required parameter processing, completeness of exception mechanism, and closed-loop of business logic. Potential hidden dangers are checked in advance to ensure that the access quality meets the standards.

Sample Conversations:

- Help me check if there are any access hidden dangers in the code, focusing on the exception handling part
- I have completed the integration, help me do a comprehensive access quality evaluation and give optimization suggestions
- Check whether the parameter processing of the settlement interface is standardized and if any required items are missing

## Docs for AI

We have optimized all technical documents to be AI-friendly, ensuring that Large Language Models (LLMs) can efficiently read, understand and flexibly use the document content, helping you get the best experience in AI programming assistants or custom Agents. The specific methods are as follows:

- Copy for LLM: The document is provided in pure Markdown format, which can be directly copied and pasted into LLM without additional format conversion, to quickly obtain accurate interpretation and usage guidance.
- llms.txt Index: The /llms.txt file is an emerging standard that helps LLMs quickly locate the core content of documents, sort out the knowledge structure, and improve retrieval and understanding efficiency.

> 💡 Supplementary Note: NPI GIT provides Newland partners and customers with the latest DEMO resources to facilitate more efficient terminal application development. To register an NPI GIT account, please contact the NPI technical support staff.