AIAny
AI Agent2026
Icon for item

herdr

Runs and monitors AI agents inside real terminal panes, surfacing agent state (blocked / working / done) and keeping agents persistent across detach/reattach. Offers workspaces, tabs, panes, socket-API integrations, and a single Rust binary for macOS/Linux.

Introduction

Why this matters

As AI agents move from experiments to everyday developer workflows, the friction comes less from model access and more from orchestration: knowing which agent is blocked, which is running, and restoring long-lived agent processes across sessions. Herdr treats agents as real terminal processes — not wrapped web views — and brings agent-aware persistence directly into the terminal you already use.

What Sets It Apart
  • Real terminal panes, not rewritten views: agents run as genuine terminal processes, so you see their native output and can interact with them directly. This reduces surprises from synthetic UI layers and preserves interactive behaviors (TTY, editors, progress bars).
  • Agent-aware multiplexer: the sidebar aggregates agent state (blocked, working, done, idle) via process-name and output heuristics plus optional socket-API integrations for more robust reporting. That makes scanning many agents quick and reliable.
  • Session persistence and handoff: server/client separation lets you detach a client while agent processes keep running. Experimental handoff attempts to migrate live panes across upgrades, and named sessions isolate runtime state per workspace.
  • Lightweight, terminal-first distribution: single Rust binary (no Electron), installable via script, Homebrew or releases. Works over SSH and interoperates with tmux, making it suitable for remote development and servers.
Who it's for & tradeoffs

Great fit if you: maintain multiple long-running agents or agent workflows and want to keep them visible and persistent inside your terminal; prefer direct terminal interaction over GUI dashboards; need simple remote/SSH attach and detach semantics without an Electron wrapper.

Look elsewhere if you: want a visual web dashboard with rich visualizations, user-friendly onboarding for non-terminal users, or need guaranteed cross-platform GUI clients (herdr targets macOS/Linux terminals and power users). The project relies on heuristics for agent state detection by default; for mission-critical orchestration, the socket-API integrations are recommended for explicit state reporting.

Where it fits

Herdr sits between classic terminal multiplexers (tmux) and GUI agent managers: it keeps the terminal-first workflows and persistence of tmux while adding agent-awareness and integrations usually found in web dashboards. For teams that already use terminals and SSH-centric workflows, it reduces context switching and preserves native CLI behaviors.

Practical notes

Expect a small runtime footprint and minimal dependencies, but plan to opt into official integrations for the most reliable agent state restoration. The dual AGPL/commercial licensing may influence adoption in some organizations; review licensing if your environment cannot comply with AGPL terms.

Information

  • Websitegithub.com
  • Authorsogulcancelik
  • Published date2026/03/27

More Items

Turns fragile, implicit search progress into explicit, persistent, shared state for multi-agent information seeking — externalizes progress as Frontier Task, Evidence Graph, Coverage Map and Failure Memory, and uses pipeline-parallel scheduling plus a middleware harness to avoid repeated failed searches and improve utilization and throughput.

GitHub
AI Agent2026

Provides a lightweight Python harness that turns LLMs into working agents with tool-use, skills, persistent memory, permission controls and multi-agent coordination. Ships with a CLI/React TUI, 43+ built-in tools, a plugin/skill system and the ohmo personal-agent for chat gateways. Best for developers prototyping agent workflows and multi-agent experiments.

GitHub
AI Model2026

Runs the Bonsai family of quantized LLMs locally (including vision-capable 27B): provides scripts and demo UIs to run 1-bit and ternary Bonsai models on macOS (Metal), Linux/Windows (CUDA/Vulkan/ROCm), or CPU, with long context, tool-calling and an optional Open WebUI agent demo.