Skip to main content
Kilo Code is an open source AI coding agent for VS Code. It generates code from natural language, automates tasks, runs terminal commands, and supports inline autocomplete.
Setup time: ~2 minutes

​
Prerequisites

​
Configuration

​
Step 1: Open Kilo Code Settings

  1. Open VS Code
  2. Click the Kilo Code icon in the sidebar
  3. Open the settings panel

​
Step 2: Select Provider

  1. In the API Provider dropdown, select OpenAI Compatible
  2. Enter the base URL: https://api.infercom.ai/v1
  3. Enter your Infercom API key
  4. Enter the model name: MiniMax-M2.5
Kilo Code API Provider Configuration

​
Step 3: Model Configuration

In the Model Configuration section, set:
SettingValue
Context Window163840
Max Output Tokens16384
Image SupportDisabled
Kilo Code Model Settings
Image Support must be disabled - MiniMax-M2.5 doesn’t support image input.

​
Step 4: Verify Connection

Start a conversation to verify the setup works.

​
Model

Use MiniMax-M2.5 - optimized for agentic coding with 160K context and 75.8% SWE-bench.

​
Usage

With Kilo Code configured:
  1. Click the Kilo Code icon in VS Code sidebar
  2. Select a mode: Architect, Coder, or Debugger
  3. Type your request in the chat
  4. Kilo Code will autonomously:
    • Read relevant files
    • Write or edit code
    • Run terminal commands
    • Iterate until the task is complete

​
Modes

ModeBest For
ArchitectPlanning, design decisions
CoderWriting and editing code
DebuggerFinding and fixing bugs

​
Example Tasks

  • “Add error handling to the login function”
  • “Write unit tests for the User class”
  • “Refactor this file to use async/await”

​
Troubleshooting

​
Connection Failed

Verify your configuration:
  • Base URL: https://api.infercom.ai/v1
  • API key is valid
  • Model name is exact: MiniMax-M2.5

​
Invalid API Key Error

Double-check your Infercom API key is entered correctly. Test it: 
curl -s https://api.infercom.ai/v1/models \
  -H "Authorization: Bearer your-infercom-api-key"

​
Model Not Found

Ensure the model name matches exactly (case-sensitive): MiniMax-M2.5

​
Slow Responses

MiniMax-M2.5 runs at 400+ tokens/sec. If responses seem slow:
  1. Check your network connection
  2. Large context (many files) increases processing time
  3. First request may be slower due to model loading

​
Next Steps