openteams-lab/openteams

Plan, Build, and Ship — with a team of AI agents instead of one

What it solves

Developers often use multiple AI coding agents (like Claude Code or Gemini CLI) in separate terminal windows, leading to fragmented context, scattered changes, and difficulty tracking project progress and costs. openteams provides a unified, local-first workspace where these agents can collaborate in a shared session, allowing developers to maintain control over the roadmap while organizing agent output.

How it works

openteams acts as a coordination layer around existing coding agents. It integrates multiple agents into a single shared session with shared context, allowing them to hand off work to one another. For complex tasks, it uses a "Workflow mode" to break down execution into structured steps and dependencies. To prevent agents from interfering with each other, it utilizes Git worktrees to create isolated workspaces for parallel tasks, which the developer can then review and merge.

Who it’s for

  • Indie developers who use multiple AI agents to build software faster.
  • Technical leads who require agent executions to be reviewable, reproducible, and organized.

Highlights

  • Multi-agent Workspace: A shared session where different AI agents can communicate and share context.
  • Workflow Mode: Structured execution of complex tasks with the ability to review, retry, or interrupt specific steps.
  • Issue Management: Developer-controlled work items that link specific tasks to agent sessions.
  • Isolated Workspaces: Use of Git worktrees to keep parallel agent tasks separate until they are ready to be merged.
  • Build Statistics: Tracking of delivered features and bug fixes against token usage and cost breakdowns.
  • Local-First: A desktop app (built with Tauri and Rust) that keeps runtime records locally.