HAR
Run coding agents with control and confidence.
HAR is an open source framework for orchestrating multi agent AI coding workflows. It lets teams run multiple coding agents in parallel, each inside an isolated environment with consistent execution, validation and teardown stages. HAR records logs, artifacts and verifiable evidence, making results easier to review and trust. It works with Claude Code, Cursor, Codex and MCP agents, and includes a CLI, MCP server and local observability dashboard.