Conductor is a multi-agent orchestration harness developed by Melty Labs. It is designed to manage, coordinate, and execute parallel coding agent workflows across local environments and cloud-based execution sandboxes. In day-to-day operations, Conductor serves as a control plane for AI-assisted development. Rather than executing a single agent prompt in a static shell, developers can dispatch tasks across multiple underlying runtimes—such as Claude Code, OpenCode, Codex, or Cursor-based workflows—using their own API keys and accounts. Workspaces can run locally on macOS or inside isolated Firecracker microVMs in Conductor Cloud, allowing agents to execute shell commands, run test suites, and prepare code diffs independently of the local machine. What sets Conductor apart from standalone terminal agents is its orchestration model and collaborative interface. It supports multi-agent workflows where different models can critique or build upon each other's outputs, alongside multiplayer workspace sharing (⌘⇧C) so multiple developers can supervise or prompt agents simultaneously within the same session. Conductor is commercial software offering a macOS desktop application, Conductor Cloud sandboxes, and an automation API, operating on a Bring Your Own Keys (BYOK) and bring-your-own-subscription model for model consumption.
Multi-Agent Orchestration — Coordinate multiple underlying agent engines such as Claude Code, OpenCode, Codex, and Cursor within a single environment. - Firecracker MicroVM Sandboxes — Spin up cloud execution environments in milliseconds with pre-installed repositories and dependencies for persistent, unattended runs. - Real-Time Multiplayer Workspaces — Share workspace links with team members to observe agent executions and co-author prompts collaboratively. - Bring Your Own Keys (BYOK) — Connect existing LLM provider subscriptions and API keys directly to route tasks to preferred models. - Model Context Protocol (MCP) — Integrate external developer tools and data sources through native Conductor MCP support. - Conductor API — Trigger headless agent sessions, dispatch tasks, and automate repository operations programmatically.
Use cases
Parallel Multi-Model Code Audits Developers can spin up concurrent agent sessions using different underlying models (e.g., asking one agent to analyze architectural risks while another writes unit tests) and consolidate their outputs into a single pull request.
Resource-Heavy Agent Execution in Cloud Sandboxes Teams offload long-running build, test, and refactoring cycles into Conductor Cloud Firecracker microVMs, preventing local machine resource exhaustion and keeping tasks running even when the laptop is closed.
Collaborative Team Agent Sessions Engineers can share a live workspace session URL with teammates to jointly prompt an agent, debug a failing production test case, and verify code diffs together before merging.
Who it is for
Conductor is built for professional software engineers, tech leads, and engineering teams using macOS who want to run multiple coding agents in parallel or offload agent tasks to cloud microVMs. It is not intended for developers looking for a minimalist, single-file terminal CLI or those working exclusively in non-macOS local desktop environments.
Install guide
Installation Download and install the native macOS application from the official site: bash # Download the Conductor desktop app directly from the official portal open https://conductor.build
First Run 1. Launch the Conductor desktop application. 2. Open settings to configure your credentials: add API keys or authenticate existing accounts (e.g., Anthropic, OpenAI, or Cursor). 3. Select your execution mode: local Mac environment or Conductor Cloud Firecracker microVM.
Example Session Within the workspace chat interface, assign tasks and spin up collaborative agent runs: text # In Conductor workspace prompt: Start a new cloud session for repo 'backend-api'. Run tests, identify flaky test cases in the auth module, refactor them using table-driven tests, and prepare a branch diff.
Okto Nexus — Okto Nexus is a local-first multi-agent coordination harness and orchestrator developed by OktoLabs. It acts as an execution bus designed…
Superset — Superset is an open-source agent orchestration harness and multi-agent workspace developed by the superset-sh team. It provides a structured environment…
Ona — Ona (formerly Gitpod's agentic platform, now part of OpenAI) is an enterprise-grade cloud agent orchestration platform that manages and runs…
Defract — Defract is an agent orchestration harness and desktop workflow environment designed to coordinate multi-stage AI software engineering workflows. Built as…
Bwee — Bwee is a native macOS agent harness designed to organize terminal-based AI agents and render interactive user interfaces on the…
Kiro — Kiro is an agentic engineering platform and CLI coding agent built and operated by AWS. It extends beyond standard code-completion…