@Alex Gleason how do your agents handle the fact that OpenCode is interactive? Do you configure OpenCode to be non-interactive with flags/settings to auto-accept all prompts or do your agents simulate human interaction (read OpenCode prompts and automatically respond)?