Parallel AI Workspace
Unscored tasks with hints and instant feedback. Practice results never become credentials.
Short lesson
Name each agent by role and task, isolate them in worktrees, hand off through files, review every diff, never paste secrets. Tool notes (2026-10): Claude Code (CLAUDE.md, agents folder), Codex CLI and others (AGENTS.md), Gemini CLI (GEMINI.md), Cursor, Copilot — verify names in each tool's docs.
Start a tmux session named pair with 2 panes and run `mock-agent A` in the first pane and `mock-agent B` in the second. Then detach and reattach.
Sign up to practiseCreate a session named agents with a window named work and two panes titled research and writer. Run `mock-agent A` in research. Copy the line containing “RESULT A” into the writer pane (via copy mode or capture-pane + buffer). Zoom the writer pane.
Sign up to practiseCreate a session named review, make a 2x2 layout, start mock agents A to D (one per pane, `mock-agent A` …), title each pane agent-a … agent-d, zoom into pane 3 (agent-c), copy one result line from pane 1 into pane 4, detach, and re-attach after a simulated connection drop (use the “Simulate disconnect” button). Save a script ~/layouts/review.sh that recreates the layout with titles.
Sign up to practiseWrite ~/.tmux.conf with prefix C-a, mouse on, vi copy-mode keys, and pane titles shown in the border (pane-border-status top). Load it into the running server.
Sign up to practiseIn ~/shop create four git worktrees (../shop-a … ../shop-d, branches agent/a … agent/d). Write a tmuxinator project ~/.config/tmuxinator/fleet.yml (session fleet) with one pane per worktree that cd's into it and runs the matching mock agent. Start it with one command: tmuxinator start fleet.
Sign up to practiseAdd a third window named monitor to session fleet with two panes: one tails ~/.agent-logs (any agent log) and one shows `agent status`. Synchronize-panes must stay off.
Sign up to practise