Nezha: Run multiple AI coding agents across projects (Claude Code and Codex)

My primary coding tools are Claude Code and Codex. I really like the CLI workflow, but managing multiple sessions in the terminal gets painful — especially when you want to revisit past conversations or switch contexts across projects.

At the same time, with AI doing most of the code generation, I’ve found myself using traditional IDEs less and less. Most of the time I just open them to read code.

So I built Nezha.

Nezha is a lightweight, agent-first coding tool that lets you run multiple AI coding agents (Claude Code / Codex) in parallel across different projects.

It combines:

multi-project workspace task / todo tracking a virtual terminal with native Claude Code / Codex integration a lightweight code editor built-in Git workflow

The idea is simple: let AI handle code generation, while humans focus on task planning and review.

You can run multiple vibe coding tasks at once, switch between them instantly, and revisit full session history when needed.

The name “Nezha” comes from a Chinese mythological character with three heads and six arms — which felt like a good metaphor for parallel AI workflows.

It’s open source, would love feedback.

One thing I noticed was that as soon as I had more than 2–3 agents running, I started losing track of what each one was doing.

Curious if others are hitting the same problem or if this is just me.