Choose a local or Docker installation for Shennian
Start with what you already have. Connect working, signed-in computer AI tools locally. Consider Docker if you have compatible API credentials and want a separately managed agent environment. Both can coexist as separate machines in Shennian.
Choose before installing
Step 1: Decide what to reuse
Choose local to reuse an existing AI login without a compatible API key. Docker uses API authentication matching the target tool and does not copy host login directories. The current Linux amd64 image provides Pi, Codex CLI, Claude Code and OpenCode. Check the latest guide for exact requirements.
Step 2: Specify directories and runtime requirements
Local execution follows the selected AI tool’s permissions. Docker needs a working environment, dedicated state storage and explicit work-directory mounts. Container AI can use mounted files and credentials. Do not mount all of HOME or the Docker socket. Keep credentials in local secret configuration, not AI chat. Containers help manage environments; they do not make every action safe.
Step 3: Follow the selected setup
Send this to your computer AI and fill in your choice:
Read https://shennian.net/install.md and reuse my existing Shennian and AI environment. I choose “local / Docker / both”. Follow the latest guide for that setup. Mount only directories I explicitly choose. I will configure authentication locally; do not read credentials or send them into chat. Finally check that the selected AI actually replies.
Approve account access yourself. Shennian is free at this stage; third-party model, API, subscription and usage rules still apply.