How to Install Kilo Code and Ship Your First AI-Assisted Feature
Step-by-step tutorial: install the Kilo Code extension, connect your favorite model provider, and use Plan Mode plus agentic mode to plan, implement, and commit a real feature.
🚀 Want to deploy Kilo Code yourself?
Docker configs, system requirements, and installation guides — all on one page.
View Kilo Code Tool Page →Step 1 — Install the extension
- Open VS Code and go to the Extensions view (
Ctrl+Shift+X). - Search for "Kilo Code" and click Install — it works in VS Code, JetBrains IDEs, and as a CLI.
- Reload the window. You will see the Kilo Code panel appear in the sidebar.
Step 2 — Choose a model provider
Kilo is provider-agnostic. In the panel settings you can pick from Anthropic Claude, Google Gemini, OpenAI ChatGPT, or a custom endpoint such as a locally hosted LLM.
| Provider | Best for |
|---|---|
| Claude (Anthropic) | Long agentic coding sessions and complex refactors. |
| Gemini (Google) | Large-context tasks with big codebases. |
| ChatGPT (OpenAI) | General-purpose development assistance. |
| Custom / local | Private, self-hosted setups via API endpoints. |
🔑 Tip: You only need an API key from your provider. Kilo never proxies your code through a closed service — prompts go straight to the model you selected.
Step 3 — Plan first, then build
- Describe a task, e.g. "Add a dark-mode toggle that persists in localStorage."
- Switch on Plan Mode: Kilo will read your project and propose a concrete implementation plan.
- Approve the plan and switch to agentic mode. Kilo edits the relevant files, checks for errors, and even runs the tests.
- Review the diff, then let it create the Git commit with a clear message.
Step 4 — Extend with MCP and rules
Attach MCP servers to give the agent real tools (databases, browsers, APIs), and add project rules so every generated change follows your team's conventions. This is where Kilo stops feeling like a chatbot and starts feeling like an engineer on your team.
Final thoughts
Kilo Code installs in minutes and scales from a personal assistant to a team-wide agentic workflow. Because it is MIT-licensed and available as an official Docker image, you can even run it in your own infrastructure. Open-source engineering has rarely felt this fast.
🚀 Deploy Kilo Code in minutes
Docker configs, system requirements, and installation guides — all on one page.
View Kilo Code Tool Page →