Hey @theo@maria_rcks@jullerino I extended T3 Code’s new usage analytics to support OpenCode alongside Codex and Claude Code.
It reads OpenCode’s SQLite history read-only, supports its legacy JSON store, and includes focused tests plus UI screenshots.
https://t.co/V3UzJGOj69
@theo My T3 Code code instance runs on my machine when I'm out for work or hangout and quite a lot of time I want to access it. Used the codex one but it sucks a lot.
@claudeai@CodexLog@opencode The Evaluator doesn't just read code — it clicks through your UI with Playwright, hits real API endpoints, runs actual CLI commands. Hard pass/fail per criterion.
https://t.co/rSPs5VpRv8
I don't let my AI write code alone anymore. Neither should you.
Planner → Generator → Evaluator. One bootstrap command. Works with @claudeai Code, @CodexLog CLI, and @opencode .