Skip to main content

Plugin for VSCode (Kilo Code)

Installation and Setup Guide for Kilo Code in VSCode​

Kilo Code is an AI programming assistant that works as a VSCode plugin. Unlike Cline, Kilo Code supports detailed cost tracking (input/output and cache pricing), even when connected to BotHub through a custom provider. The assistant helps create and edit files, execute commands, and more, requesting permission at every step.

How to Get Started with Kilo Code?​

  1. Launch VS Code on your computer. 1.1 If VS Code shows the warning "Running extensions might...", click the Allow button.

  2. Download Kilo Code. 2.1 Open the Extensions panel (icon on the sidebar or Cmd+Shift+X (macOS) / Ctrl+Shift+X (Windows/Linux)). 2.2 Type Kilo Code into the search bar. 2.3 Find Kilo Code in the extension results and click Install.

We recommend creating a separate folder for Kilo Code projects:

  • macOS: Documents/Kilo-Code
  • Windows: Documents/Kilo-Code
  • Inside, you can create subfolders for each project.
  • Open the created folder in VSCode through the File -> Open Folder menu.
  1. Open Kilo Code. 3.1 The icon will appear in the activity bar. Click it. If the icon does not appear, restart VS Code. 3.1.1 Another way to open it -> command palette. Cmd+Shift+P (macOS) / Ctrl+Shift+P (Windows/Linux), then run the command Kilo Code: Open In New Tab.

Configuring Connections to BotHub​

When launching for the first time, you will need to add BotHub as a custom provider.

  1. Adding a provider. 1.1 Click the gear icon in the plugin window to open settings. 1.2 Go to the Providers section. 1.3 Add a Custom provider.

Important about the API key. Some versions of Kilo Code have bugs related to saving the key in the API Key field. Leave this field empty. The key must be specified in the request headers.

  1. Configure parameters. 2.1 Name: Enter any name, for example, BotHub. 2.2 BaseURL: https://openai.bothub.chat/v1 2.3 Headers: Click Add Header and create the Authorization field. In the value, enter Bearer <Key>, replacing <Key> with your API key generated on the page https://bothub.chat/profile/for-developers. 2.4 API Key: Leave empty.

Configuring Cost Tracking​

For Kilo Code to correctly calculate token costs, you need to manually specify model prices and parameters in the configuration file.

  1. Open the settings file. 1.1 Click the gear icon in the plugin window again. 1.2 Find the button to open the global or local configuration file.

  2. Fill in models and prices: 2.1 If the provider has already been added, find the section with its name and extend the models list. 2.2 If the provider has not been added yet, paste the block below into the root of the JSON file, replacing <YOUR_API_KEY> and specifying актуальные models. 2.3 Always check the current models and prices in the table on the page https://bothub.chat/models.

Below are example pricing configurations for the "settings.json" file:

  "$schema": "https://app.kilo.ai/config.json",
  "permission": {
    "bash": "allow"
  },
  "disabled_providers": [],
  "provider": {
    "bothub-igm": {
      "name": "BotHub IGM",
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://openai.bothub.chat/v1",
        "headers": {
          "Authorization": "Bearer <YOUR_API_KEY>"
        }
      },
      "models": {
        "gpt-5.4": {
          "name": "GPT-5.4",
          "reasoning": false,
          "supportsCaching": true,
          "cost": {
            "input": 405,
            "output": 2430,
            "cache_read": 40.5,
            "cache_write": 405
          },
          "contextLength": 1050000,
          "maxCompletionTokens": 128000
        },
        "gpt-5.5": {
          "name": "GPT-5.5",
          "reasoning": true,
          "supportsCaching": true,
          "cost": {
            "input": 810,
            "output": 4860,
            "cache_read": 81,
            "cache_write": 810
          },
          "contextLength": 1050000,
          "maxCompletionTokens": 128000
        },
        "deepseek-v4-pro": {
          "name": "DeepSeek V4 Pro",
          "reasoning": false,
          "supportsCaching": true,
          "cost": {
            "input": 70.47,
            "output": 140.94,
            "cache_read": 7.05,
            "cache_write": 70.47
          },
          "contextLength": 1048576,
          "maxCompletionTokens": 384000
        },
        "deepseek-v4-flash": {
          "name": "DeepSeek V4 Flash",
          "reasoning": false,
          "supportsCaching": true,
          "cost": {
            "input": 22.68,
            "output": 45.36,
            "cache_read": 2.27,
            "cache_write": 22.68
          },
          "contextLength": 1048576,
          "maxCompletionTokens": 384000
        },
        "gpt-4.1-mini": {
          "name": "GPT-4.1 Mini",
          "reasoning": false,
          "supportsCaching": true,
          "cost": {
            "input": 64.8,
            "output": 259.2,
            "cache_read": 6.48,
            "cache_write": 64.8
          },
          "contextLength": 1047576,
          "maxCompletionTokens": 32768
        }
      }
    }
  }
}```

## Cache Cost Calculation
In the example above, cache prices are defined according to the principle: `cache_write` = `input`, `cache_read` = `0.1 of input`.

## Getting Started

After configuring everything, give it a test task. For example:

"Create a folder named test in the directory and write a small Python script for a simple calculator (addition, subtraction, division, multiplication) that works in the terminal."

Press `Enter`, and Kilo Code will start execution while asking for your permission before performing actions.