We built *Hyve* — a CLI tool that lets you run multiple AI coding agents in *true isolated workspaces*.
Most tools let agents generate code, but as soon as you run two at once you hit a wall:
- git conflicts (stash/checkout pain)
- shared DB state breaking tests
- port collisions between services
- environment mixups (wrong .env per agent)
With Hyve each agent gets:
separate git worktrees per feature
isolated Postgres (or dev DB) per workspace
auto-configured ports + env variables
reliable service orchestration + health checks
This means you can run:
Agent A exploring new feature
hyve create feat-A server webapp
hyve run feat-A
Agent B fixing bugs
hyve create bugfix -B server webapp
hyve run bugfix -B
pgsql
Copy code
And both environments run side-by-side with zero conflicts.
It works great with Claude Code, Cursor, Aider, or any multi-agent workflow.
We built *Hyve* — a CLI tool that lets you run multiple AI coding agents in *true isolated workspaces*.
Most tools let agents generate code, but as soon as you run two at once you hit a wall:
- git conflicts (stash/checkout pain) - shared DB state breaking tests - port collisions between services - environment mixups (wrong .env per agent)
With Hyve each agent gets:
separate git worktrees per feature isolated Postgres (or dev DB) per workspace auto-configured ports + env variables reliable service orchestration + health checks
This means you can run:
Agent A exploring new feature hyve create feat-A server webapp hyve run feat-A
Agent B fixing bugs hyve create bugfix -B server webapp hyve run bugfix -B
pgsql Copy code
And both environments run side-by-side with zero conflicts. It works great with Claude Code, Cursor, Aider, or any multi-agent workflow.
*Install & try:* https://github.com/eladkishon/hyve
We’re happy to answer questions or hear feedback!