tmux

Use tmux when a task needs a persistent terminal state across tool calls.

Prerequisites

bash

tmux -V

If tmux is not installed, stop and ask the user to install it.

Why tmux for Agents

tmux sessions survive SSH drops, terminal closes, and laptop sleep. This makes them ideal for AI agent workflows:

•Persistence — a long-running build or test suite continues even if your terminal disconnects.
•Detach/reattach — reconnect to a session hours later and see full scrollback.
•Isolation — each agent gets its own session with independent state.
•Observability — attach from another terminal to watch an agent work in real time.

Socket Convention

Use a dedicated socket directory:

bash

SOCKET_DIR="${VIBES_TMUX_SOCKET_DIR:-${TMPDIR:-/tmp}/vibes-tmux-sockets}"
mkdir -p "$SOCKET_DIR"
SOCKET="$SOCKET_DIR/vibes.sock"

Use short session names with no spaces. Target panes by window name (e.g. session:shell) rather than numeric index (e.g. session:0.0) — numeric indexes depend on base-index/pane-base-index settings and are not portable.

Quickstart

bash

SESSION="vibes-work"
tmux -S "$SOCKET" new-session -d -s "$SESSION" -n shell
tmux -S "$SOCKET" send-keys -t "$SESSION":shell -l -- "echo ready"
sleep 0.1
tmux -S "$SOCKET" send-keys -t "$SESSION":shell Enter
tmux -S "$SOCKET" capture-pane -p -J -t "$SESSION":shell -S -200

After creating a session, show monitor commands:

bash

tmux -S "$SOCKET" attach -t "$SESSION"
tmux -S "$SOCKET" capture-pane -p -J -t "$SESSION":shell -S -200

Sending Input Reliably

•Use literal sends for text payloads: send-keys -l -- "$cmd".
•Send Enter separately with a short delay.
•Send control keys separately (for example C-c).

bash

tmux -S "$SOCKET" send-keys -t "$SESSION":shell -l -- "$cmd"
sleep 0.1
tmux -S "$SOCKET" send-keys -t "$SESSION":shell Enter

Do not combine text and Enter in one fast send for interactive agent TUIs.

Live Supervision Workflow

Use this when a user wants to watch an agent run and intervene live.

bash

# start session and launch agent
tmux -S "$SOCKET" new-session -d -s review-agent -n shell
tmux -S "$SOCKET" send-keys -t review-agent:shell -l -- "cd /path/to/repo && codex"
sleep 0.1
tmux -S "$SOCKET" send-keys -t review-agent:shell Enter

# watch live in another terminal
tmux -S "$SOCKET" attach -t review-agent

# from operator terminal: inspect and steer
tmux -S "$SOCKET" capture-pane -p -J -t review-agent:shell -S -200
tmux -S "$SOCKET" send-keys -t review-agent:shell -l -- "run tests and summarize failures"
sleep 0.1
tmux -S "$SOCKET" send-keys -t review-agent:shell Enter
tmux -S "$SOCKET" send-keys -t review-agent:shell C-c

Use explicit session names per task (for example review-agent, fix-auth, release-check) so operators can track runs quickly.

Common Commands

bash

# list sessions and panes
tmux -S "$SOCKET" list-sessions
tmux -S "$SOCKET" list-panes -a

# capture output
tmux -S "$SOCKET" capture-pane -p -J -t "$SESSION":shell -S -400

# interrupt running command
tmux -S "$SOCKET" send-keys -t "$SESSION":shell C-c

Agent CLI Patterns

Use one session per tool or repo for isolation.

bash

# Claude Code
tmux -S "$SOCKET" new-session -d -s claude-main -n shell
tmux -S "$SOCKET" send-keys -t claude-main:shell -l -- "cd /path/to/repo && claude"
sleep 0.1
tmux -S "$SOCKET" send-keys -t claude-main:shell Enter

# Codex
tmux -S "$SOCKET" new-session -d -s codex-main -n shell
tmux -S "$SOCKET" send-keys -t codex-main:shell -l -- "cd /path/to/repo && codex"
sleep 0.1
tmux -S "$SOCKET" send-keys -t codex-main:shell Enter

# GitHub Copilot CLI (detect installed command)
COPILOT_CMD=""
for cmd in copilot ghcs; do
  command -v "$cmd" >/dev/null 2>&1 && COPILOT_CMD="$cmd" && break
done
if [[ -z "$COPILOT_CMD" ]] && gh copilot --help >/dev/null 2>&1; then
  COPILOT_CMD="gh copilot"
fi

if [[ -n "$COPILOT_CMD" ]]; then
  tmux -S "$SOCKET" new-session -d -s copilot-main -n shell
  tmux -S "$SOCKET" send-keys -t copilot-main:shell -l -- "cd /path/to/repo && $COPILOT_CMD"
  sleep 0.1
  tmux -S "$SOCKET" send-keys -t copilot-main:shell Enter
fi

When prompting these tools later, keep using the same session and split text/Enter.

Git Worktrees for Agent Isolation

Use git worktree to give each agent its own working directory without cloning:

bash

# create worktrees from the main repo
git worktree add ../repo-agent-a -b agent-a
git worktree add ../repo-agent-b -b agent-b

# one tmux session per worktree
tmux -S "$SOCKET" new-session -d -s agent-a -n shell
tmux -S "$SOCKET" send-keys -t agent-a:shell -l -- "cd ../repo-agent-a"
sleep 0.1
tmux -S "$SOCKET" send-keys -t agent-a:shell Enter

tmux -S "$SOCKET" new-session -d -s agent-b -n shell
tmux -S "$SOCKET" send-keys -t agent-b:shell -l -- "cd ../repo-agent-b"
sleep 0.1
tmux -S "$SOCKET" send-keys -t agent-b:shell Enter

Each agent works on an independent branch with no lock contention on the index.

Multi-Agent Orchestration

Run multiple agents in parallel, each in its own session and worktree:

bash

AGENTS=("frontend" "backend" "tests")
for agent in "${AGENTS[@]}"; do
  # create worktree if it doesn't exist
  [[ -d "../repo-$agent" ]] || git worktree add "../repo-$agent" -b "$agent"

  # create session
  tmux -S "$SOCKET" new-session -d -s "$agent" -n shell
  tmux -S "$SOCKET" send-keys -t "$agent":shell -l -- "cd ../repo-$agent"
  sleep 0.1
  tmux -S "$SOCKET" send-keys -t "$agent":shell Enter
done

# monitor all sessions
tmux -S "$SOCKET" list-sessions

# check output from a specific agent
tmux -S "$SOCKET" capture-pane -p -J -t frontend:shell -S -200

Helpers

Use bundled scripts:

•scripts/find-sessions.sh to inspect sessions/sockets.
•scripts/wait-for-text.sh to wait for a prompt or ready marker.

Both scripts support -S (socket path) and -L (socket name) flags.

Examples:

bash

{baseDir}/scripts/find-sessions.sh -S "$SOCKET"
{baseDir}/scripts/wait-for-text.sh -S "$SOCKET" -t "$SESSION":shell -p 'ready|Done|❯|\\$'

Cleanup

bash

tmux -S "$SOCKET" kill-session -t "$SESSION"
tmux -S "$SOCKET" kill-server
# clean up worktrees when done
git worktree remove ../repo-agent-a

References

For tmux fundamentals beyond this workflow guide, see references/fundamentals.md:

•Session/window/pane structure and targeting
•Window and pane management (splits, resize)
•Advanced capture-pane and scrollback configuration
•Environment variables and troubleshooting

Notes

•tmux supports macOS/Linux natively. On Windows, use WSL.
•Prefer tmux for interactive tools, auth flows, and REPLs.
•Prefer normal exec/background jobs for non-interactive commands.