OpenCode Runner

OpenCode Runner

This runner integrates with the OpenCode CLI. Shipped in Untether v0.5.0.

Installation

npm i -g opencode-ai@latest

Configuration

Add to your untether.toml:

=== “untether config”

```sh
untether config set opencode.model "claude-sonnet"
```

=== “toml”

```toml
[opencode]
model = "claude-sonnet"  # optional
```

Usage

untether opencode

Resume Format

Resume line format: `opencode --session ses_XXX`

The runner recognizes both --session and -s flags (with or without run).

Note: The resume line is meant to reopen the interactive TUI session. opencode run is headless and requires a message or command, so it is not the canonical resume command.

JSON Event Format

OpenCode outputs JSON events with the following types:

Event Type	Description
`step_start`	Beginning of a processing step
`tool_use`	Tool invocation with input/output
`text`	Text output from the model
`step_finish`	End of a step (reason: “stop” or “tool-calls” when present)
`error`	Error event

See stream-json-cheatsheet.md for detailed event format documentation.

Known Limitations

No auto-compaction

OpenCode does not support automatic context compaction. Unlike Pi (which emits AutoCompactionStart/AutoCompactionEnd events to trim context) and Claude Code (which manages its context window internally), OpenCode sessions accumulate unbounded context across turns.

Impact: Long sessions with many prompts will progressively slow down as the full conversation history is sent to the model on every turn. A session that starts at 72k tokens can grow past 77k+ after just 4-5 prompts.

Workaround: Start a fresh session with /new when response times degrade noticeably.

If OpenCode adds compaction events in the future, Untether will need schema and runner updates following the Pi compaction pattern.