Lore / Origin

Why this fork exists

I tried OpenCode because the CLI experience was solid, but a few things didn’t click for me:

• The sidebar wasn’t collapsible
• Selection always copied everything
• I couldn’t connect multiple coding plan providers (Codex, Gemini, Antigravity)
• And most importantly, I couldn’t see coding‑plan usage in one place

So I forked it.

The goal

I use Codex, Z.ai, Claude Code, Gemini, and Antigravity. I wanted a single CLI that lets me:

• run all of them from one place
• see 5‑hour and weekly limits together
• keep a consistent workflow regardless of provider

The benchmark idea

I still want a clean, fair way to compare models — the same task, same constraints, different agents. Benchmarking is a core goal, but it isn’t exposed as a stable CLI command yet. The plan is to make it a reliable way to test “can this model ship features?” without losing the ability to undo or re‑apply changes.