Summary

Maestro transforms the Gemini CLI into a sophisticated multi-agent development orchestration platform. It addresses complex software development tasks by delegating work to 22 specialized AI agents, coordinated by a central "TechLead" orchestrator. This provides engineers and power users with a structured, AI-driven approach to design, planning, execution, and review, significantly enhancing productivity and code quality.

How It Works

Maestro employs a "TechLead" orchestrator to manage 22 distinct subagents, each with specialized expertise. For simple tasks, an "Express" workflow offers a streamlined, single-agent delegation. Medium to complex tasks utilize a structured 4-phase "Standard" workflow: Design Dialogue, Planning, Execution, and Completion. A key innovation is native parallel subagent execution, allowing independent phases to run concurrently. Session state is persistently stored in YAML+Markdown files, enabling reliable resumption after interruptions.

Quick Start & Requirements

• Prerequisites: Gemini CLI must have its experimental subagent system enabled. Manually set "experimental": {"enableAgents": true} in ~/.gemini/settings.json.
• Installation: Install via Gemini CLI: gemini extensions install https://github.com/josstei/maestro-gemini. For local development, clone the repository and use gemini extensions link ..
• Verification: Run gemini extensions list after installation.
• Primary Command: /maestro:orchestrate <task description> initiates a full orchestration.
• Documentation: The README serves as the primary documentation.

Highlighted Details

• 22 Specialized Agents: Covers 8 domains including Engineering, Product, Design, SEO, Compliance, and more.
• Dual Workflow System: Offers a rapid "Express" mode for simple tasks and a comprehensive 4-phase "Standard" workflow for complex projects.
• Native Parallel Execution: Leverages Gemini CLI's scheduler for concurrent processing of independent implementation phases.
• Persistent Sessions: Tracks all orchestration state in YAML+Markdown files, allowing for robust interruption recovery.
• Standalone Tools: Provides direct access to specialized agents for code review, debugging, security audits, performance analysis, SEO, accessibility, and compliance checks.
• Least-Privilege Security: Agents are granted only necessary tools via a 4-tier access model.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or roadmap were found within the provided README text.

Licensing & Compatibility

• License: Apache-2.0.
• Compatibility: The Apache-2.0 license is permissive and generally compatible with commercial use and linking in closed-source projects.

Limitations & Caveats

Native parallel subagents currently operate in an "autonomous mode (YOLO)," implying auto-approval of tool calls. The system requires enabling experimental features in the Gemini CLI. Potential issues with overlapping file ownership during parallel execution are noted, and users may need to manually resolve conflicts or switch to sequential mode.