What problem does it solve? Coordinating multiple coding agents, shells, and background commands across terminals is error-prone without a stable control surface. This Skill lets an agent inspect and control Herdr-managed workspaces, tabs, and panes through the herdr CLI, so it can spawn agents, send prompts, read output, and wait for status changes reliably. ## Core Features & Use Cases - Pane-based agent control: Split panes, launch interactive agents (Codex, Claude Code, pi, OpenCode, OMP), submit prompts, and read transcripts using stable pane IDs. - State-aware waiting: Wait on agent status transitions (idle, working, blocked, done) or output matches before reading results, with timeout handling. - Topology and safety rules: Discover workspaces, tabs, and worktrees via JSON responses, keep user focus intact with --no-focus, and avoid destructive operations on sessions you did not create. - Use Case: While working in a Herdr session, ask the agent to spin up a reviewer agent in a sibling pane, wait for it to finish in the background, and report its findings without leaving your current pane. ## Quick Start Ask the agent to use Herdr to open a reviewer agent in a new pane and have it review the current diff.
