Your coding agent resends files, diffs, instructions, and logs it has already seen. Halv cleans that wasted context locally before each request reaches the model—so you keep the same CLI, login, model, and subscriptions while getting far more work from your plan. It supports Claude Code, Codex, Kimi, and GLM, with split terminals, multi-agent workflows, local sandboxing, a Crux code index, and a receipt showing the usage saved on every answer. Available on macOS, Windows, and Linux.

Hey Product Hunt — Pedro here, maker of Halv 👋 I built Halv after noticing that coding agents resend the same files, diffs, instructions, and logs on every turn. That repeated context burns through paid plan limits without improving the answer. Halv removes the redundant context locally before the request reaches the model. You keep the same CLI, login, model, and workflow. Every answer includes a receipt showing what it cost and what it would have cost without Halv. Halv also brings Claude Code, Codex, Kimi, and GLM into one multi-agent workspace with split terminals, local sandboxing, subagent tracking, and the Crux code index. It is free to test, with no card or API key required. I would love your feedback—especially from developers who regularly hit AI coding plan limits.

Halv

Make your AI coding plan last twice as long.

Make your AI coding plan last twice as long.