Kilo Code

Kilo Code — Documentation & Integration Guide

Editor & Environment Setup

Kilo Code is available as an extension for VS Code and all major JetBrains IDEs (IntelliJ IDEA, PyCharm, WebStorm), as a CLI tool, and as Cloud Agents for headless or CI/CD workflows.

Configuring AI Inference

Developers can connect Kilo Code to AI models via four pathways:

Method	Description
Free Models	Zero-cost inference via supported free-tier model providers
BYOK	Supply your own API keys for Anthropic, OpenAI, Google, Azure, AWS Bedrock, Ollama, LM Studio, and 60+ other providers
Kilo Gateway	Managed proxy to 500+ models across 60+ providers at direct provider rates, no markup
Kilo Pass	Subscription from $19/mo; includes up to 50% bonus credits

Auto Model Routing can be enabled to let Kilo Code select the optimal model per task based on cost and capability.

Agent Modes

• Code — direct code generation and editing
• Architect — high-level design and planning
• Debug — root-cause analysis and fix suggestions
• Ask — codebase Q&A and explanation
• Custom — user-defined agent behavior

Cloud Agents & Parallelism

Cloud Agents run tasks in parallel isolated worktrees, enabling non-blocking concurrent execution across multiple tasks or branches.

Integrations

• Version control & project management: Bitbucket, Jira
• Communication: Slack
• Local model runtimes: Ollama, LM Studio
• Protocol extensibility: MCP (Model Context Protocol) for custom context injection

Self-Hosting

Kilo Code is MIT-licensed and self-hostable, suitable for air-gapped or data-residency-constrained environments. Source available on GitHub (26,700+ stars).