A developer built an AI-powered homelab management system using OpenCode Web UI with Git integration. The setup pushes changes through pull requests for manual approval before GitOps deploys updates via Arcane for Docker services and other tools. This approach eliminates hours of manual release note checking and container updates.

The platform runs on a VM with the OpenCode webserver as a systemd unit, featuring built-in terminal, file browser, and git worktree support. OpenCode operates under its own user account with dedicated SSH keys, preventing direct pushes to production branches. This workflow maintains security while enabling AI-assisted code modifications across multiple devices.

Previously, updating a dozen Docker Compose stacks required manual review of release notes and breaking changes. Now the AI summarizes release notes and adds healthchecks to containers, reducing update time from hours to minutes. The developer migrated services from Truenas to Arcane GitOps projects for better version control.

The main limitation involves CI feedback, as Forgejo Actions lacks public API access to job logs unlike GitHub Actions. This prevents the AI from diagnosing failing tests or stack traces automatically. Despite this gap, the platform provides secure, auditable AI-assisted infrastructure management with minimal attack surface.