Usage Documentation

XSTRIVE LLM API Management Platform — User Guide

This document guides you through the complete usage flow from unboxing to getting started

---

Table of Contents

1. Hardware Guide
2. Software Guide
3. Visual Studio Code setup
4. FAQ

---

I. Hardware Instructions

1.1 Packing List

After opening the package, please confirm that the following items are complete:

📌 Note: All products only provide an electronic manual. Please keep this document safe or visit the official website to download the latest version.

---

1.2 First Use

Step 1: Connect the Device

1. Plug the network cable into the device network port
2. Connect the power adapter; the device powers on automatically
3. Make sure the computer and the device are connected to the same switch / router.

┌─────────────┐      网线      ┌─────────────┐
│   路由器    │◄──────────────►│  迅思维盒子  │
│  /交换机    │                │             │
└──────┬──────┘                └─────────────┘
       │
       │ 同一网络
       ▼
┌─────────────┐
│    电脑     │
└─────────────┘

Step 2: search and configure device IP

1. Download the "Device Search Assistant"
   - Download address: http://www.xstrive.com/download/

2. Operation device search assistant, automatic search for rapid thinking devices in local area networks

3. If the device is not found, tick "Force search" and search again

4. When the device is found, the device IP address can be modified (recommended to be set to a fixed IP to facilitate subsequent access)

---

1.3 Login Management Interface

This device is configured through a web management interface and supports the following terminals:

• ✅ PC (recommended)
• ✅ Tablet
• ✅ Phone

💡 On a PC, Google Chrome is recommended for the best compatibility.

Login step:

1. Open Google Chrome
2. Enter the device IP address in the address bar (e.g.http://192.168.0.24)
3. Open the XSTRIVE LLM API Management Platform sign-in page

---

II. Software instructions

2.1 Sign-in

⚠️ Security hint: Change the default password immediately after the first login

Enter the main interface of the platform when login is successful. The top menu bar contains the following functional modules:

[Channels], [Tokens], [Redeem], [Top up], [Users], [Overview], [Logs], [Settings], [Network], [About]

---

2.2 Network Configuration

Configure network connectivity parameters to ensure that the equipment is stable and accessible.

Path: top menu → [Network]

Field descriptions

📌 Recommended : disable DHCP in production and use a static IP so that an IP change cannot interrupt the service. The device supports several network ports (port 0 / port 1) and you can switch between them as needed.

---

2.3 Channel management

A channel is the route to an upstream AI model provider. Once a channel is configured the platform can proxy calls to the major AI models.

Path: top menu → [Channels]

Channel List

• Displays all allocated channels in table form, including: ID, name, grouping, type, status, response time, balance
• Status indicator — 🟢 green "Enabled" means online and available, 🔴 means disabled
• Quick actions — each channel supports "Test" (check connectivity), "Delete", "Disable", "Edit"
• Bulk actions — supports testing all channels, testing disabled channels, deleting disabled channels, refreshing the list
• Top notice — OpenAI channels no longer support obtaining balance through the key; manual refresh is required

Add Channel

Click the " Add new channel" button to open the creation page:

Required fields:

Optional fields:

• Model redirect — JSON format, used to modify the model name mapping in the request body
• System prompt — Set a default system prompt for this channel
• Proxy — Configure the proxy server address (format: https://domain.com)
• Batch creation — tick this to create several channels at once

💡 Recommendation: Configure at least 2 channels to open load balance and enhance service stability. API Key Retrieval address:

• OpenAI:https://platform.openai.com/
• Claude:https://console.anthropic.com/
• ChatGLM:https://open.bigmodel.cn/
• DeepSeek:https://platform.deepseek.com/

---

2.4 Token management

A token is the credential for calling the API. A token controls which IPs may connect, which models they may reach and how much quota they may spend.

Path: top menu → [Tokens]

Token list

• Table displays all tokens, including: name, status, amount used, remaining amount, creation time, expiry time
• Status indicator — 🟢 green "Enabled" means valid, 🔴 means disabled or expired
• Quick actions — each token supports "Copy Key", "Disable", "Delete", "Enable" and "Edit"
• Sorting — supports sorting by multiple dimensions

Create token

Click the " Add new token" button to open the creation page:

Configure fields:

⚠️ Note: the token Key is shown only once, when the token is created, so keep it safe. If you lose it you have to create a new token. The token quota only caps what this token may spend; actual usage is also limited by the remaining balance of the account.

---

2.5 Redemption code management

Path: top menu → [Redeem]

• Redemption code list — a table of all redemption codes, with ID, name, status (unused / used), face value, creation time and redemption time
• Search & filter — supports quick retrieval by redemption code ID and name
• Row actions — each record supports "Copy", "Delete", "Enable / disable" and "Edit"
• Create & refresh — add a new redemption code with one click, or refresh the list to get the latest status

💡 Typical uses: education and training, employee purchase schemes, user incentives and more. Codes can be generated in bulk and handed out by email or QR code.

---

2.6 Top-up centre

Path: top menu → [Top up]

• Account balance — the current balance is shown in large type at the top (denominated in USD)
• Get a redemption code — click the button to get a new redemption code for distribution or personal use
• Redeem a code — after entering the redemption code, click "Redeem Now" and the denomination is automatically credited to the account balance

⚠️ The top-up centre is for ordinary users. Administrators who need to generate redemption codes in bulk should use the "Redemption code management" module.

---

2.7 User Management

Path: top menu → [Users]

• User list — all users displayed in a table, including username, group, total quota, used quota, request count, role, activation status
• Role hierarchy — three-level role permissions:
• 🟢 User — can only call the API and view their own data
• 🟡 Administrator - Manages user, token, channel etc. operations
• 🟠 Super administrator — holds every permission in the system (including settings and network configuration)
• Quick actions — each user supports "Promote / demote role", "Delete account", "Disable / enable" and "Edit"
• Add user — administrators can manually create new user accounts

---

2.8 Overview dashboard

Path: top menu → [Overview]

• Model request trend — a line chart showing the call volume of each model over time, identifying popular models and usage peaks
• Quota consumption trend — a line chart showing daily quota consumption, assisting cost forecasting and budget control
• Token consumption trend — a line chart showing the token usage trend, reflecting the actual scale of AI resource consumption
• The last 7 days are shown by default; the charts update in real time

---

2.9 Logs

Path: top menu → [Logs]

• Multi-dimension filters — supports combined filtering by token name, model name, time range, channel ID, username and other conditions
• Log detail table — each record includes timestamp, channel number, type, model name, username, token name, Prompt Token, Completion Token, deducted quota, and detail expansion
• Test replay — you can directly initiate a "Test" on historical logs to reproduce that request
• Full-text search — log content supports keyword search to quickly locate abnormal requests

💡 Logs are the only basis for billing reconciliation. Export a backup regularly and keep it for at least 90 days.

---

2.10 Personal Settings

Path: top menu → [Settings] → Personal Settings

• General settings — explains what the system access token is used for
• Quick actions — update your profile, generate a system access token, copy the invitation link or delete your personal account
• Account binding — bind or unbind an email address

---

2.11 Operating Settings

Path: top menu → [Settings] → Operating Settings

• Quota settings — new user initial quota, per-request deduction quota, invitation reward quota, etc.
• Rate settings — core pricing engine:
• Model rate — set a price rate for each model individually
• Completion rate — additional rate adjustment for the completion stage
• Group rate — different user groups enjoy different price rates
• Preset rates — mainstream model rates (GPT series, Claude, DeepSeek, ChatGLM, etc.) are built in, ready to use out of the box

---

2.12 System Settings

Path: top menu → [Settings] → System Settings

• Server address — set the system public network access address, which affects API endpoints and callback addresses
• Login & registration — control the registration entry (password login / registration switch / email verification switch)
• Email domain allow-list — restrict registration to specific email domains only
• SMTP mail service — configure the mail server for sending verification codes and notifications

---

III. Using Visual Studio Code

Below is a description of how to configure the LLM API platform in VS Code as an AI programming assistant.

3.1 Open the chat panel

1. Open VS Code
2. Click the Chat icon in the left activity bar (or press Ctrl+Alt+I)

3.2 Add a custom model

1. Open the “Model” drop-down and choose “Select Model”
   
2. Choose “Other Models” and click “Manage Language Models”
   

3.3 Configure the model

1. Click “Add Model”
   
2. Choose “Custom”
   
3. Choose a Group (you can create one, e.g. "XSTRIVE")
   
4. Enter the Key you copied in token management
   

3.4 Fill in configuration information

🔗 URL building rules (important)

http://{设备IP}:3000/v1/chat/completions?key={你的令牌Key}

Example:

http://192.168.0.24:3000/v1/chat/completions?key=sk-DqobQ4SlBtcKQjft4445D2A6F43b498fA67d9fBd5bFc258c

💬 Questions?: URL building is a little involved, so feel free to contact support for one-to-one guidance.

3.5 Getting started

Once configured, pick the model you just added in the VS Code chat panel and start AI-assisted coding.

---

IV. FAQ

Q1: the device search assistant cannot search the device

Solution:

1. Make sure the computer and the device are on the same network (the same switch / router)
2. Check that the device is powered on and its network cable is connected properly
3. Try ticking the "Force search" option
4. Temporarily disable the computer's firewall or antivirus software and try again

Q2: the page looks wrong after signing in?

Solution:

• Switch to Google Chrome to open the page
• Clear the browser cache and refresh the page
• Check whether the device IP has changed (e.g. when DHCP is in use)

Q3: channel test fails?

Solution:

1. Check that the API Key is correct (watch for stray spaces)
2. Confirm the Key is valid and has balance on the provider's platform
3. Check that the network can reach the provider's API address
4. If a proxy is in use, check that the proxy configuration is correct

Q4: The model does not respond in VS Code?

Solution:

1. Check that the URL is built correctly (especially the IP and the Key)
2. Confirm the token has not expired and its quota is not used up
3. Check that VS Code can reach the device IP (same network)
4. Check the XSTRIVE platform logs to troubleshoot