OpenCode 2.0 Launches with Breaking Plugin API & Runtime Overhaul

Background: OpenCode 1.x Architecture

OpenCode 1.x was built on Bun as the core runtime for fast startup and Tauri for a minimal desktop footprint. These choices reflected early open‑source “romanticism” prioritizing raw speed and tiny install size.

Why the Rewrite: From Bun/Tauri to Node.js/Electron

As daily active users grew to millions, the team hit ecosystem and stability limits. Bun, while fast, lacked the depth and maturity of Node.js tooling. Tauri struggled with cross‑OS stability under heavy front‑end rendering and long‑running processes. The pragmatic decision was to switch back to Node.js and Electron, accepting larger binaries and slightly slower startup in exchange for system‑level stability and a foundation for complex commercial scenarios.

Plugin System Overhaul

The plugin system is now central to OpenCode 2.0. It adopts an everything is a plugin philosophy (possibly inspired by DSH) with hot‑reload support. Internally, over 80 plugins already cover most of the new version’s functionality.

Three Breaking Changes

Plugin API overhaul : 1.x plugins cannot run; they must be rewritten against the new interfaces.

Server API redesign : Any external scripts calling the old server endpoints must be completely rewritten.

Configuration merge : Terminal settings previously split in tui.json are now unified into global cli.json (automatic migration provided).

Community Reaction

The long beta period caused confusion; some developers complained about missing release notes and broken custom scripts. Others noted that moving sessions to a persistent local background service — shared across TUI, web, and desktop — is a necessary step toward “service‑oriented” AI coding tools. The author acknowledges that a minority of users may leave due to the scale of changes, but views the “rip‑off‑the‑band‑aid” approach as paying down technical debt from rapid early growth.

Upgrade Process and Licensing

Core CLI and desktop remain MIT licensed . To prevent silent breakage, 2.0 updates are no longer auto‑installed; users must manually run opencode upgrade --major.

Conclusion

The painful migration is framed as laying a stronger foundation for a more robust AI programming ecosystem. Despite short‑term friction, the benefits of a stable, extensible architecture are expected to outweigh the costs.