Documentation

Kaiden Docs

Kaiden is an open desktop platform for running AI coding agents in isolated, secured environments. Every agent gets its own sandbox — it can only reach what you've allowed, touch the files you've shared, and call the models you've configured.

Get started First sandbox →

Getting Started

◈

What is Kaiden?

The core idea: sandboxes, security, and why every agent needs its own isolated environment.

⬡

Your First Sandbox

Walk through the five-step wizard to create a secured sandbox for any repository.

Configuration

◉

Credentials & Secrets

Store API keys and tokens in the Secret Vault. Agents never see the values directly.

◎

AI Agents

Configure Claude Code, Codex, Goose, and other agents with per-agent defaults.

◇

Models & Inference

Cloud providers, local Ollama models, enterprise OpenShift AI, and semantic routing.

Organizing Work

▣

Projects

Save credentials, network rules, and skills as a project so every workspace inherits them.

▤

Skills, MCP & Knowledge

Give agents domain skills, live tool access via MCP servers, and retrieval over your docs.

Security & Reference

◆

Network Security

Deny-by-default network policy, reading blocked events, and live policy updates.

◈

Settings

OpenShell runtime, CLI tools, experimental features, and the status bar reference.