OpenParable
Skip to content
← Back to skills

Parallel Processing

Productivity
parallel-processing

Parallel Processing is an Agent Skill for productivity published in the majiayu000/claude-skill-registry repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Run multiple tasks in parallel using tmux windows or Agent Teams

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/majiayu000/claude-skill-registry/blob/HEAD/skills/orchestration/parallel-processing-kamushadenes-nix-2/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/parallel-processing, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Parallel Processing

Use tmux for background tasks, Agent Teams for multi-instance parallel work.

Agent Teams (Preferred for Multi-Task Work)

For 2+ independent tasks that each need a full Claude session:

Create an Agent Team with one teammate per task.
Each teammate gets a spawn prompt with their task details.
Teammates can discuss findings across domains.

Use /delegate-tasks to automate team creation for task-master tasks.

Recommended: Use Subagents for Context Efficiency

For simple parallel commands, launch multiple tmux-runner subagents in parallel:

# Launch all three in parallel (single message with multiple Task calls)
Task(subagent_type="tmux-runner", description="Run tests",
     prompt="Run 'npm test' and return results", model="haiku")
Task(subagent_type="tmux-runner", description="Run lint",
     prompt="Run 'npm run lint' and return results", model="haiku")
Task(subagent_type="tmux-runner", description="Run build",
     prompt="Run 'npm run build' and return results", model="haiku")

This saves context in the main conversation - you only see the results, not all the intermediate tmux operations.

Pattern: Background Task (Direct)

For long-running servers you need to monitor:

# Start background task
window_id = tmux_new_window(command="npm run dev", name="dev-server")

# Continue working...

# Check later
output = tmux_capture(target=window_id, lines=50)

# Clean up
tmux_kill(target=window_id)

Pattern: With Notification

window_id = tmux_new_window(command="npm run build", name="build")
status = tmux_wait_idle(target=window_id, timeout=300)
notify(title="Build Complete", message=f"Status: {status}")

Pattern: Multiple Parallel (Direct)

When you need to manage windows yourself:

test_win = tmux_new_window(command="npm test", name="tests")
lint_win = tmux_new_window(command="npm run lint", name="lint")
build_win = tmux_new_window(command="npm run build", name="build")

for w in [test_win, lint_win, build_win]:
    tmux_wait_idle(target=w, timeout=120)