execution

Orchestrate Maestro workflow phase execution with parallel or sequential delegation.

453|34|Updated Feb 9, 2026
One-click install
npx skills add https://github.com/josstei/maestro-gemini --skill execution-josstei
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: execution
Source: https://github.com/josstei/maestro-gemini/tree/main/skills/execution
Command: npx skills add https://github.com/josstei/maestro-gemini --skill execution-josstei

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Maestro's Execution phase provides a governed mechanism to run phases safely, choosing between parallel and sequential delegation and enforcing gating until mode is established.

Core Features & Use Cases

  • Execution mode gate that blocks delegation until MAESTRO_EXECUTION_MODE is set.
  • Supports parallel and sequential execution with batch control using MAESTRO_MAX_CONCURRENT.
  • Implements sequential protocol steps: in_progress, current_phase, current_batch, and write_todos before delegation.
  • Provides robust error handling, state persistence, and retry logic to recover from failures.
  • Integrates with hooks and session management to archive results upon completion.

Quick Start

Set MAESTRO_EXECUTION_MODE to parallel or sequential, then trigger the Execution phase to delegate to native subagents with proper session state updates.

Frequently Asked Questions about execution

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate parallel and sequential phase execution in Maestro workflows?

Orchestrate Maestro phase execution by setting the MAESTRO_EXECUTION_MODE variable to parallel or sequential. This gates delegation to native subagents, ensuring safe and predictable execution with proper session state updates.

How does the execution mode gate work for subagent delegation?

The execution mode gate blocks subagent delegation until MAESTRO_EXECUTION_MODE is explicitly set to parallel or sequential. This mechanism ensures predictable execution by preventing unconfigured tasks from running before the workflow mode is established.

How do I control batch concurrency during parallel execution?

Control batch concurrency during parallel execution by configuring the MAESTRO_MAX_CONCURRENT variable. This limits the number of native subagents running simultaneously, preventing resource exhaustion and governing delegation safely.

Can I recover from subagent failures during sequential execution?

Sequential execution supports robust error handling and retry logic to recover from subagent failures. It persists session state using protocol steps like in_progress, current_phase, and current_batch to ensure workflow continuity after errors.

What is the best way to manage session state when delegating tasks to subagents?

Manage session state by following the sequential protocol steps: set in_progress, current_phase, current_batch, and write_todos before delegation. This ensures state persistence and integrates with hooks to archive results upon completion.

Why does my Maestro workflow block before delegating to subagents?

Your workflow blocks because the execution mode gate prevents delegation until MAESTRO_EXECUTION_MODE is configured. This safety mechanism ensures phase execution only proceeds once parallel or sequential mode is explicitly established.