Kilo Code

Use TokenMix models inside an IDE coding agent.

Prepare TokenMix values

• Provider: OpenAI Compatible
• Base URL: https://api.tokenmix.ai/v1
• API Key: your TokenMix key
• Model: a TokenMix model ID such as
• For code indexing embeddings: prepare an embedding model ID and dimension, such as 1536 or 3072, based on the TokenMix model page.

Configure the VS Code extension

1. Install Kilo Code from the VS Code Marketplace.
2. Open Kilo Code settings or model configuration.
3. Choose OpenAI Compatible.
4. Paste your TokenMix API Key.
5. Set Base URL to https://api.tokenmix.ai/v1.
6. Set Model to a TokenMix model ID.
7. Keep advanced options such as reasoning, tool use, and Responses API unchanged until basic chat works.

CLI setup

Install the CLI:

npm install -g @kilocode/cli

Run kilo inside a project directory. If the CLI asks for a provider, choose OpenAI Compatible and enter the same TokenMix Base URL, API Key, and Model.

Test with small tasks

Start with small tasks, not a whole-project refactor:

Create a README section explaining how to run this project locally.

Then test code generation:

Create a TypeScript helper that formats a USD price to 6 decimals and trims trailing zeros.

Troubleshooting

• 401: wrong API key.
• 404/model not found: wrong TokenMix API model ID.
• tool calls fail: verify normal chat first, then enable file-editing and terminal permissions.
• code indexing fails: configure embedding provider as OpenAI-compatible with TokenMix too.
• auto mode risk: test CLI auto/CI mode in a disposable repo first.

Extra notes for the TokenMix app guide

Kilo Code has IDE, CLI, and platform entry points, so the guide should avoid mixing them. In the app center, lead with the VS Code extension path and keep CLI as an optional developer path.

Recommended beginner path

1. Install the VS Code extension first, not the CLI.
2. Configure OpenAI Compatible with TokenMix.
3. Use Ask mode to explain a file and confirm context reading works.
4. Then allow it to write one small file.
5. Only after that, test terminal commands, browser automation, or MCP.

Code indexing note

If users enable codebase indexing, they also need embeddings. Kilo docs list OpenAI-Compatible embeddings with baseUrl and apiKey. Use https://api.tokenmix.ai/v1 and the TokenMix key. The embedding dimension should match the TokenMix embedding model documentation.