Headless OpenCode tasks in your coding workspace

Use OpenCode for configured headless work while keeping its queue and results visible beside the rest of your coding tasks.

Configure the provider before dispatch

Install OpenCode separately and configure at least one model provider. VibeiDE uses that configuration when it starts a bounded headless execution. It does not supply a new model subscription or turn an unconfigured CLI into a working provider.

Give headless work a task lifecycle

Choose the repository, write the request, and set the project’s OpenCode capacity. The task waits, runs, and records its outcome under the same queue lifecycle used by other providers. Its activity and result remain available for review after the process ends.

Use the right interface expectation

OpenCode execution here is headless. It does not expose the interactive terminal continuation available for supported Codex and Claude Code sessions. Plan council and the Standup workflow use Codex and Claude. Choose OpenCode for work that fits its supported execution path.

Review before moving to the next dependency

Documentation checks, a bounded investigation, or an independent batch change can be suitable prompts when your configured model supports them. Verify the result against the actual request. A completed process is evidence of an outcome to inspect, not a guarantee that the change is ready to ship.

Go deeper

Languages