Terminal multiplexer for coordinating multiple AI coding agents
Orchestrate multiple AI coding agents across persistent sessions, a real-time status sidebar, and mouse-resizable panes. Herdr Extended, from Herdr contributors, runs on Windows as a lightweight Rust-built utility and targets developers who run concurrent agent workflows. It bundles persistent background servers, SSH multi-machine support, and a local socket API for programmatic pane control, aiming to keep long-running automated tasks and their state intact when switching machines or reconnecting.
How it organises multi-agent workflows
The app acts as a centralized runtime for several AI coding agents by tracking agent lifecycles and grouping terminals into workspaces. Agent State Tracking marks agents as "working", "blocked", "done", or "idle", and Workspace Organization maps tabs and panes to projects or repositories. These elements let developers keep agent context tied to a project while switching between tasks on the same desktop or across machines.
What it requires and how it runs on systems
The tool runs as a single Rust binary with no external dependencies, noted as a roughly 10MB download, which simplifies deployment on desktop environments. Persistent Sessions are provided by a background server that keeps panes and processes alive across client restarts or disconnections. Multi-Machine Support uses SSH so you can reattach to the same live session from another device.
Is it safe and scriptable for automated agents
Herdr Extended exposes a local Socket API that lets agents spawn panes and send input programmatically, which supports coordinated automation across agents. The project is open source under the Apache License 2.0, allowing code inspection and verification of server behaviour. The background server model preserves running processes, which reduces accidental termination when a client disconnects.
Who can operate it and the learning curve
The mouse-native interface supports clicking, dragging to resize panes, and right-click menus, while keyboard shortcuts remain available. Casual users can use the GUI elements, whereas teams that integrate agents programmatically will need familiarity with the Socket API and SSH session management. The developer-focused design targets engineers who run multiple long-lived agent tasks rather than casual single-shell users.
Clear choice for developers managing persistent AI agent fleets, with a narrow focus
The app suits developers who need durable, project-linked agent sessions and programmatic control across machines. Its lightweight, single-binary design and agent-aware monitoring favour long-running automated workflows. The main trade-off is its specialised focus on AI agent orchestration rather than general-purpose terminal multiplexing, so users who need a generic CLI-only multiplexer may find its scope narrower.
