
from skills274
Control the herdr terminal-native agent multiplexer: list workspaces/tabs/panes, run commands, read output, and wait for agent or output state changes over a lo
The herdr skill documents how an agent can control and observe a running herdr instance — a terminal-native multiplexer that exposes workspaces, tabs, and panes over a local unix socket. This capability lets agents orchestrate parallel tasks, spawn new agent panes, run commands in isolated shells, read live terminal output, and wait for specific output or agent-status transitions. The skill explains pane/workspace concepts, safe interaction patterns (do not redirect pane output), and concrete CLI recipes for common flows (run a server and wait for readiness, run tests in a separate pane, spawn an agent and coordinate multi-turn conversations).
Use when the agent is executing inside a herdr environment (HERDR_ENV=1) and needs to manage multiple terminal contexts or interact with other agents and processes live. Ideal for running builds, starting servers, watching logs, running tests, coordinating other agent instances, and automating terminal-first developer workflows. Not applicable outside a herdr runtime.
herdr wait output, herdr wait agent-status), and best practices for pacing and observing output.Best with agents that run in CLI environments and can execute subprocesses or shell commands (Codex/Copilot-like environments, CLI-focused agent runtimes).
This skill has not been reviewed by our automated audit pipeline yet.