Orchestrate AI coding agents in parallel, each in its own git worktree. Review diffs, comment inline, and send feedback to resume — from one desktop IDE.

Launch agents across repos from a single panel. Track live status and diff stats at a glance — get notified the moment an agent needs your input.

Name your task, pick an agent, choose a branch — and go. Deep integration with Claude Code, plus support for any CLI agent you can run in a terminal.

Every agent runs in its own git worktree — a real branch with full history. No file conflicts, no stepping on each other. Review the diff, commit, and merge when you're ready.

Write comments directly on diffs — offline or pulled from GitHub. Select the ones that matter and send them back to the agent in one click. The agent resumes with your full context.

Syntax highlighting, go-to-definition, and autocomplete powered by Language Server Protocol. Plus an integrated terminal for when you need to step in yourself.

An open-source desktop IDE for orchestrating parallel AI coding agents. Define tasks, let agents run in isolated git worktrees simultaneously, then review and iterate from a single interface.
Workstreams is agent-agnostic. Claude Code has deep integration with session capture, resume, and auto-commit. Aider and Gemini are also supported, with Codex coming soon.
Each agent gets its own isolated git worktree — a separate working directory sharing the same .git history. Agents work in parallel without file conflicts. Worktree creation is serialized to prevent git lock races, then agents run fully parallel.
Yes. Workstreams is fully open source. It orchestrates AI agents that you already have installed — there are no extra accounts, subscriptions, or API keys required by Workstreams itself.
Workstreams is a standalone desktop IDE for macOS with full LSP support, syntax highlighting, and an integrated terminal. It runs as its own app — no extensions or plugins needed.
Yes. Workstreams supports pulling online comments from GitHub alongside local offline review comments. Both are included when resuming agents, so your feedback loop spans across tools.
Free and open source. No accounts, no API keys.