DuDuClaw OS Docs
A Linux where AI coworkers are native residents. Start here to install it, set it up, and understand what runs underneath.
Start with these
Section titled “Start with these”Quick startDownload an image, verify it with minisign, write it to a USB stick, boot.
Desktop editionWhat the desktop looks like, how the agent works on its shadow workspace and freezes the moment a human touches the keyboard or mouse, lock screen and first-run setup.
Hardware requirementsThe x86-64-v3 / UEFI / SSD hard floor, recommended mini PCs, and the known driver gaps.
App compatibilitycompat.d runners, Windows apps through Bottles, Android through Waydroid — and what is not promised.
AI runtimesWhich AI CLIs ship in the image, where credentials live, how to regenerate the runtime bundle.
Keyboard shortcutsGlobal compositor bindings, the shell UI, and Cmd+K to hand over a task at any moment.
Local modelsRunning a model on the box with the bundled llama.cpp llama-server: offline inference, routing, hardware floor.
Building the OSBuild an image from source with kas, boot it in QEMU, then sign and publish.
Reporting a security issueDisclosure channel, scope, and how to verify release artifacts.
Sections
Section titled “Sections”FeaturesEvery capability from the user's side, starting with the appliance; goal loop, memory, channels and the evolution engine live in this section.
GuidesDeployment, development, Docker, channel wiring, evals — steps that actually run.
ArchitectureThe system overview and the design of the evolution engine.
SpecificationsThe open formats: SOUL.md, CONTRACT.toml, SKILL.md, expert packs.
APIGateway protocol and the OpenAPI reference.
← Back to the DuDuClaw OS site · DuDuClaw platform site · GitHub