OpenCode 2.0 Released, Letting Models Write Scripts to Orchestrate Tools

AI coding tool OpenCode 2.0 has officially launched, with the headline capability being that models can write their own scripts to orchestrate tools, according to open-source community outlet OSCHINA.

The key phrase in the report is that models can write scripts to dispatch tools. Under that description, a model is no longer limited to a single call against one individual tool; it can generate a script for a task and let that script chain several tools into a repeatable workflow.

That capability is worth isolating because the bottleneck for coding agents is rarely whether they can write code at all. It is whether they can carry a sequence of actions through to the end reliably, and a script is one way to lock a multi-step procedure into something that runs the same way every time.

OpenCode sits in the category of AI coding tools, and the 2.0 release pushes its capability boundary further toward autonomous execution.

Letting a model write the scripts that orchestrate tools also raises new questions: how much permission a script holds, where it is allowed to run, and how a bad or over-reaching execution gets rolled back. Those boundaries decide whether such a tool can be trusted inside real repositories and production environments.

What to watch next is the concrete implementation detail behind the 2.0 release, how the developer community actually adopts it, and whether conventions emerge around sandboxing and permissions for model-written scripts.