Runs several coding agents in parallel on separate branches, then lets you compare results.
Conductor runs several coding agents in parallel, each on its own branch and workspace, then lets you review and compare what they produced. Instead of waiting on one agent, you send the same task three ways and keep the best result. It is infrastructure for a working style that only makes sense once agents are fast and cheap.
- Parallel agents on separate branches
- Isolated workspace per agent
- Compare results side by side
- Same task, several attempts
- Review before merging
- Local git worktrees per run
Recent X Posts
Introducing the Conductor MCP! Give this to your agent to get started: ``` Add the Conductor MCP server at api.conductor.build/mcp and authenticate with OAuth. ``` x.com/charlieholtz/s…
three things we love: Trains, Orchestras, and Paper!! x.com/paper/status/2…
just one more thing x.com/charlieholtz/s…