GitHub - pedramamini/Maestro: Agent Orchestration Command Center

Maestro

Run AI coding agents autonomously for days.

Maestro is a cross-platform desktop app for orchestrating your fleet of AI agents and projects. It’s a high-velocity solution for hackers who are juggling multiple projects in parallel. Designed for power users who live on the keyboard and rarely touch the mouse.

Collaborate with AI to create detailed specification documents, then let Auto Run execute them automatically, each task in a fresh session with clean context. Allowing for long-running unattended sessions, my current record is nearly 24 hours of continuous runtime.

Run multiple agents in parallel with a Linear/Superhuman-level responsive interface. Currently supporting Claude Code, OpenAI Codex, and OpenCode with plans for additional agentic coding too…

Maestro

Run AI coding agents autonomously for days.

Installation

Download

Download the latest release for your platform from the Releases page:

macOS: .dmg or .zip
Windows: .exe installer
Linux: .AppImage, .deb, or .rpm
Upgrading: Simply replace the old binary with the new one. All your data (sessions, settings, playbooks, history) persists in your config directory.

Requirements

At least one supported AI coding agent installed and authenticated:
Claude Code - Anthropic’s AI coding assistant
OpenAI Codex - OpenAI’s coding agent
OpenCode - Open-source AI coding assistant
Git (optional, for git-aware features)

Features

Power Features

🤖 Auto Run & Playbooks - File-system-based task runner that batch-processes markdown checklists through AI agents. Create playbooks for repeatable workflows, run in loops, and track progress with full history. Each task gets its own AI session for clean conversation context.
💬 Group Chat - Coordinate multiple AI agents in a single conversation. A moderator AI orchestrates discussions, routing questions to the right agents and synthesizing their responses for cross-project questions and architecture discussions.
🌐 Mobile Remote Control - Built-in web server with QR code access. Monitor and control all your agents from your phone. Supports local network access and remote tunneling via Cloudflare for access from anywhere.
💻 Command Line Interface - Full CLI (maestro-cli) for headless operation. List agents/groups, run playbooks from cron jobs or CI/CD pipelines, with human-readable or JSONL output for scripting.
🚀 Multi-Instance Management - Run unlimited agents and terminal sessions in parallel. Each agent has its own workspace, conversation history, and isolated context.
📬 Message Queueing - Queue messages while AI is busy; they’re sent automatically when the agent becomes ready. Never lose a thought.

Core Features

🔄 Dual-Mode Sessions - Each agent has both an AI Terminal and Command Terminal. Switch seamlessly between AI conversation and shell commands with Cmd+J.
⌨️ Keyboard-First Design - Full keyboard control with customizable shortcuts. Cmd+K quick actions, rapid agent switching, and focus management designed for flow state.
📋 Session Discovery - Automatically discovers and imports all Claude Code sessions, including conversations from before Maestro was installed. Browse, search, star, rename, and resume any session.
🔀 Git Integration - Automatic repo detection, branch display, diff viewer, commit logs, and git-aware file completion. Work with git without leaving the app.
📁 File Explorer - Browse project files with syntax highlighting, markdown preview, and image viewing. Reference files in prompts with @ mentions.
🔍 Powerful Output Filtering - Search and filter AI output with include/exclude modes, regex support, and per-response local filters.
⚡ Slash Commands - Extensible command system with autocomplete. Create custom commands with template variables for your workflows.
💾 Draft Auto-Save - Never lose work. Drafts are automatically saved and restored per session.
🔊 Speakable Notifications - Audio alerts with text-to-speech announcements when agents complete tasks.
🎨 Beautiful Themes - 12 themes including Dracula, Monokai, Nord, Tokyo Night, GitHub Light, and more.
💰 Cost Tracking - Real-time token usage and cost tracking per session and globally.
🏆 Achievements - Level up from Apprentice to Titan of the Baton based on cumulative Auto Run time. 11 conductor-themed ranks to unlock.

Note: Maestro supports Claude Code, OpenAI Codex, and OpenCode. Support for additional agents (Aider, Gemini CLI, Qwen3 Coder) may be added in future releases based on community demand.

Spec-Driven Workflow

Maestro enables a specification-first approach to AI-assisted development. Instead of ad-hoc prompting, you collaboratively build detailed specs with the AI, then execute them systematically:

┌─────────────────────────────────────────────────────────────────────┐
│  1. PLAN          2. SPECIFY         3. EXECUTE        4. REFINE    │
│  ─────────        ──────────         ─────────         ─────────    │
│  Discuss the      Create markdown    Auto Run works    Review       │
│  feature with     docs with task     through tasks,    results,     │
│  the AI agent     checklists in      fresh session     update specs │
│                   your Auto Run      per task          and repeat   │
│                   folder                                            │
└─────────────────────────────────────────────────────────────────────┘

Why this works:

Deliberate planning — Conversation forces you to think through requirements before coding
Documented specs — Your markdown files become living documentation
Clean execution — Each task runs in isolation with no context bleed
Iterative refinement — Review, adjust specs, re-run—specs evolve with your understanding

Example workflow:

Plan: In the AI Terminal, discuss your feature: "I want to add user authentication with OAuth support"
Specify: Ask the AI to help create a spec: "Create a markdown checklist for implementing this feature"
Save: Copy the spec to your Auto Run folder (or have the AI write it directly)
Execute: Switch to Auto Run tab, select the doc, click Run—Maestro handles the rest
Review: Check the History tab for results, refine specs as needed

This approach mirrors methodologies like Spec-Kit, but with a graphical interface, real-time AI collaboration, and multi-agent parallelism.

Key Concepts

Concept	Description
Agent	A workspace tied to a project directory and AI provider (Claude Code, Codex, or OpenCode). Contains one Command Terminal and one AI Terminal with full conversation history.
Group	Organizational container for agents. Group by project, client, or workflow.
Group Chat	Multi-agent conversation coordinated by a moderator. Ask questions across multiple agents and get synthesized answers.
AI Terminal	The conversation interface with your AI agent. Supports `@` file mentions, slash commands, and image attachments.
Command Terminal	A PTY shell session for running commands directly. Tab completion for files, git branches, and command history.
Session Explorer	Browse all past conversations for an agent. Star, rename, search, and resume any previous session.
Auto Run	Automated task runner that processes markdown checklists. Spawns fresh AI sessions per task.
Playbook	A saved Auto Run configuration with document order, options, and settings for repeatable batch workflows.
History	Timestamped log of all actions (user commands, AI responses, Auto Run completions) with session links.
Remote Control	Web interface for mobile access. Local network or remote via Cloudflare tunnel.
CLI	Headless command-line tool for scripting, automation, and CI/CD integration.

UI Overview

Maestro features a three-panel layout:

Left Panel - Agent list with grouping, filtering, search, bookmarks, and drag-and-drop organization
Main Panel - Center workspace with two modes per agent:
AI Terminal - Converse with your AI agent (Claude Code, Codex, or OpenCode). Supports multiple tabs/sessions, @ file mentions, image attachments, slash commands, and draft auto-save.
Command Terminal - PTY shell with tab completion for files, branches, tags, and command history.
Views: Session Explorer, File Preview, Git Diffs, Git Logs
Right Panel - Three tabs: File Explorer, History Viewer, and Auto Run

Agent Status Indicators

Each session shows a color-coded status indicator:

🟢 Green - Ready and waiting
🟡 Yellow - Agent is thinking
🔴 Red - No connection with agent
🟠 Pulsing Orange - Attempting to establish connection
🔴 Red Badge - Unread messages (small red dot overlapping top-right of status indicator, iPhone-style)

Screenshots

All these screenshots were captured in the them "Pedurple". For screenshots of other themes, see THEMES.md. Also note that these screenshots are probably dated as the project is evolving rapidly.

Main Screen

Command Interpreter (with collapsed left panel)

Git Logs and Diff Viewer

File Viewer

CMD+K and Shortcuts Galore

Themes and Achievements

Session Tracking, Starring, Labeling, and Recall

AutoRuns with Change History Tracking

Group Chat

Web Interface / Remote Control

Chat

Groups / Sessions

History

Keyboard Shortcuts

Global Shortcuts

Action	macOS	Windows/Linux
Quick Actions	`Cmd+K`	`Ctrl+K`
Toggle Sidebar	`Cmd+B`	`Ctrl+B`
Toggle Right Panel	`Cmd+\`	`Ctrl+\`
New Agent	`Cmd+N`	`Ctrl+N`
Kill Agent	`Cmd+Shift+Backspace`	`Ctrl+Shift+Backspace`
Move Agent to Group	`Cmd+Shift+M`	`Ctrl+Shift+M`
Previous Agent	`Cmd+[`	`Ctrl+[`
Next Agent	`Cmd+]`	`Ctrl+]`
Jump to Agent (1-9, 0=10th)	`Opt+Cmd+NUMBER`	`Alt+Ctrl+NUMBER`
Switch AI/Command Terminal	`Cmd+J`	`Ctrl+J`
Show Shortcuts Help	`Cmd+/`	`Ctrl+/`
Open Settings	`Cmd+,`	`Ctrl+,`
View All Agent Sessions	`Cmd+Shift+L`	`Ctrl+Shift+L`
Jump to Bottom	`Cmd+Shift+J`	`Ctrl+Shift+J`
Cycle Focus Areas	`Tab`	`Tab`
Cycle Focus Backwards	`Shift+Tab`	`Shift+Tab`

Panel Shortcuts

Action	macOS	Windows/Linux
Go to Files Tab	`Cmd+Shift+F`	`Ctrl+Shift+F`
Go to History Tab	`Cmd+Shift+H`	`Ctrl+Shift+H`
Go to Auto Run Tab	`Cmd+Shift+1`	`Ctrl+Shift+1`
Toggle Markdown Raw/Preview	`Cmd+E`	`Ctrl+E`
Insert Checkbox (Auto Run)	`Cmd+L`	`Ctrl+L`

Input & Output

Action	Key
Send Message	`Enter` or `Cmd+Enter` (configurable in Settings)
Multiline Input	`Shift+Enter`
Navigate Command History	`Up Arrow` while in input
Slash Commands	Type `/` to open autocomplete
Focus Output	`Esc` while in input
Focus Input	`Esc` while in output
Open Output Search	`Cmd+F` while in output
Scroll Output	`Up/Down Arrow` while in output
Page Up/Down	`Alt+Up/Down Arrow` while in output
Jump to Top/Bottom	`Cmd+Up/Down Arrow` while in output

Tab Completion (Command Terminal)

The Command Terminal provides intelligent tab completion for faster command entry:

Action	Key
Open Tab Completion	`Tab` (when there’s input text)
Navigate Suggestions	`Up/Down Arrow`
Select Suggestion	`Enter`
Cycle Filter Types	`Tab` (while dropdown is open, git repos only)
Cycle Filter Backwards	`Shift+Tab` (while dropdown is open)
Close Dropdown	`Esc`

Completion Sources:

History - Previous shell commands from your session
Files/Folders - Files and directories in your current working directory
Git Branches - Local and remote branches (git repos only)
Git Tags - Available tags (git repos only)

In git repositories, filter buttons appear in the dropdown header allowing you to filter by type (All, History, Branches, Tags, Files). Use Tab/Shift+Tab to cycle through filters or click directly.

@ File Mentions (AI Terminal)

In AI mode, use @ to reference files in your prompts:

Action	Key
Open File Picker	Type `@` followed by a search term
Navigate Suggestions	`Up/Down Arrow`
Select File	`Tab` or `Enter`
Close Dropdown	`Esc`

Example: Type @readme to see matching files, then select to insert the file reference into your prompt. The AI will have context about the referenced file.

Navigation & Search

Action	Key
Navigate Agents	`Up/Down Arrow` while in sidebar
Select Agent	`Enter` while in sidebar
Open Session Filter	`Cmd+F` while in sidebar
Navigate Files	`Up/Down Arrow` while in file tree
Open File Tree Filter	`Cmd+F` while in file tree
Open File Preview	`Enter` on selected file
Close Preview/Filter/Modal	`Esc`

File Preview

Action	macOS	Windows/Linux
Copy File Path	`Cmd+P`	`Ctrl+P`
Open Search	`Cmd+F`	`Ctrl+F`
Scroll	`Up/Down Arrow`	`Up/Down Arrow`
Close	`Esc`	`Esc`

Most shortcuts are customizable in Settings > Shortcuts

Slash Commands

Maestro includes an extensible slash command system with autocomplete. Type / in the input area to open the autocomplete menu, use arrow keys to navigate, and press Tab or Enter to select.

Custom AI Commands

Create your own slash commands in Settings > Custom AI Commands. Each command has a trigger (e.g., /deploy) and a prompt that gets sent to the AI agent.

Commands support template variables that are automatically substituted at runtime:

Agent Variables

Variable	Description
`{{AGENT_NAME}}`	Agent name
`{{AGENT_PATH}}`	Agent home directory path (full path to project)
`{{AGENT_GROUP}}`	Agent’s group name (if grouped)
`{{AGENT_SESSION_ID}}`	Agent session ID (for conversation continuity)
`{{TAB_NAME}}`	Custom tab name (alias: `SESSION_NAME`)
`{{TOOL_TYPE}}`	Agent type (claude-code, codex, opencode)

Path Variables

Variable	Description
`{{CWD}}`	Current working directory
`{{AUTORUN_FOLDER}}`	Auto Run documents folder path

Auto Run Variables

Variable	Description
`{{DOCUMENT_NAME}}`	Current Auto Run document name (without .md)
`{{DOCUMENT_PATH}}`	Full path to current Auto Run document
`{{LOOP_NUMBER}}`	Current loop iteration (starts at 1)

Date/Time Variables

Variable	Description
`{{DATE}}`	Current date (YYYY-MM-DD)
`{{TIME}}`	Current time (HH:MM:SS)
`{{DATETIME}}`	Full datetime (YYYY-MM-DD HH:MM:SS)
`{{TIMESTAMP}}`	Unix timestamp in milliseconds
`{{DATE_SHORT}}`	Short date (MM/DD/YY)
`{{TIME_SHORT}}`	Short time (HH:MM)
`{{YEAR}}`	Current year (YYYY)
`{{MONTH}}`	Current month (01-12)
`{{DAY}}`	Current day (01-31)
`{{WEEKDAY}}`	Day of week (Monday, Tuesday, etc.)

Git & Context Variables

Variable	Description
`{{GIT_BRANCH}}`	Current git branch name (requires git repo)
`{{IS_GIT_REPO}}`	"true" or "false"
`{{CONTEXT_USAGE}}`	Current context window usage percentage

Example: A custom /standup command with prompt:

It's {{WEEKDAY}}, {{DATE}}. I'm on branch {{GIT_BRANCH}} at {{AGENT_PATH}}.
Summarize what I worked on yesterday and suggest priorities for today.

Auto Run

Auto Run is a file-system-based document runner that lets you batch-process tasks using AI agents. Select a folder containing markdown documents with task checkboxes, and Maestro will work through them one by one, spawning a fresh AI session for each task.

Setting Up Auto Run

Navigate to the Auto Run tab in the right panel (Cmd+Shift+1)
Select a folder containing your markdown task documents
Each .md file becomes a selectable document

Creating Tasks

Use markdown checkboxes in your documents:

# Feature Implementation Plan

- [ ] Implement user authentication
- [ ] Add unit tests for the login flow
- [ ] Update API documentation

Tip: Press Cmd+L (Mac) or Ctrl+L (Windows/Linux) to quickly insert a new checkbox at your cursor position.

Running Single Documents

Select a document from the dropdown
Click the Run button (or the ▶ icon)
Customize the agent prompt if needed, then click Go

Multi-Document Batch Runs

Auto Run supports running multiple documents in sequence:

Click Run to open the Batch Runner Modal
Click + Add Docs to add more documents to the queue
Drag to reorder documents as needed
Configure options per document:

Reset on Completion - Uncheck all boxes when document completes (for repeatable tasks)
Duplicate - Add the same document multiple times

Enable Loop Mode to cycle back to the first document after completing the last
Click Go to start the batch run

Playbooks

Save your batch configurations for reuse:

Configure your documents, order, and options
Click Save as Playbook and enter a name
Load saved playbooks from the Load Playbook dropdown
Update or discard changes to loaded playbooks

Git Worktree Support

Git worktrees allow Auto Run to operate in an isolated directory while you continue working interactively in the main repository—no read-only restrictions, no waiting.

Setup:

Enable Worktree in the Batch Runner Modal
Select a Worktree Directory (base folder for all worktrees)
Enter a Branch Name (becomes the worktree subdirectory)
The computed path shows where the worktree will be created
Optionally enable Create PR on completion to auto-open a pull request

Benefits:

Continue working in the main repo while Auto Run processes tasks in the background
No read-only mode — the yellow border and input lock only appear when running without a worktree
Visual indicator — a git branch icon appears in the AUTO badge, right panel, and status pill when using a worktree
Automatic PR creation — when enabled, opens a PR from the worktree branch to your target branch

Without a worktree: Auto Run queues with other write operations to prevent conflicts, and the editor enters read-only mode until completion.

Progress Tracking

The runner will:

Process tasks serially from top to bottom
Skip documents with no unchecked tasks
Show progress: "Document X of Y" and "Task X of Y"
Mark tasks as complete (- [x]) when done
Log each completion to the History panel

Session Isolation

Each task executes in a completely fresh AI session with its own unique session ID. This provides:

Clean context - No conversation history bleeding between tasks
Predictable behavior - Tasks in looping playbooks execute identically each iteration
Independent execution - The agent approaches each task without memory of previous work

This isolation is critical for playbooks with Reset on Completion documents that loop indefinitely. Without it, the AI might "remember" completing a task and skip re-execution on subsequent loops.

Environment Variables {#environment-variables}

Maestro sets environment variables that your agent hooks can use to customize behavior:

Variable	Value	Description
`MAESTRO_SESSION_RESUMED`	`1`	Set when resuming an existing session (not set for new sessions)

Example: Conditional Hook Execution

Since Maestro spawns a new agent process for each message (batch mode), agent "session start" hooks will run on every turn. Use MAESTRO_SESSION_RESUMED to skip hooks on resumed sessions:

# In your agent's session start hook
[ "$MAESTRO_SESSION_RESUMED" = "1" ] && exit 0
# ... rest of your hook logic for new sessions only

This works with any agent provider (Claude Code, Codex, OpenCode) since the environment variable is set by Maestro before spawning the agent process.

History & Tracking

Each completed task is logged to the History panel with:

AUTO label indicating automated execution
Session ID pill (clickable to jump to that AI conversation)
Summary of what the agent accomplished
Full response viewable by clicking the entry

Keyboard navigation in History:

Up/Down Arrow - Navigate entries
Enter - View full response
Esc - Close detail view and return to list

Auto-Save

Documents auto-save after 5 seconds of inactivity, and immediately when switching documents. Full undo/redo support with Cmd+Z / Cmd+Shift+Z.

Image Support

Paste images directly into your documents. Images are saved to an images/ subfolder with relative paths for portability.

Stopping the Runner

Click the Stop button at any time. The runner will:

Complete the current task before stopping
Preserve all completed work
Allow you to resume later by clicking Run again

Parallel Batches

You can run separate batch processes in different Maestro sessions simultaneously. Each session maintains its own independent batch state. With Git worktrees enabled, you can work on the main branch while Auto Run operates in an isolated worktree.

Group Chat

Group Chat lets you coordinate multiple AI agents in a single conversation. A moderator AI orchestrates the discussion, routing questions to the right agents and synthesizing their responses.

When to Use Group Chat

Cross-project questions: "How does the frontend authentication relate to the backend API?"
Architecture discussions: Get perspectives from agents with different codebase contexts
Comparative analysis: "Compare the testing approach in these three repositories"
Knowledge synthesis: Combine expertise from specialized agents

How It Works

Create a Group Chat from the sidebar menu
Add participants by @mentioning agent names (e.g., @Frontend, @Backend)
Send your question - the moderator receives it first
Moderator coordinates - routes to relevant agents via @mentions
Agents respond - each agent works in their own project context
Moderator synthesizes - combines responses into a coherent answer

The Moderator’s Role

The moderator is an AI that controls the conversation flow:

Direct answers: For simple questions, the moderator responds directly
Delegation: For complex questions, @mentions the appropriate agents
Follow-up: If agent responses are incomplete, keeps asking until satisfied
Synthesis: Combines multiple agent perspectives into a final answer

The moderator won’t return to you until your question is properly answered—it will keep going back to agents as many times as needed.

Example Conversation

You: "How does @Maestro relate to @RunMaestro.ai?"

Moderator: "Let me gather information from both projects.
@Maestro @RunMaestro.ai - please explain your role in the ecosystem."

[Agents work in parallel...]

Maestro: "I'm the core Electron desktop app for AI orchestration..."

RunMaestro.ai: "I'm the marketing website and leaderboard..."

Moderator: "Here's how they relate:
- Maestro is the desktop app (the product)
- RunMaestro.ai is the website (discovery and community)
- They share theme definitions for visual consistency

Next steps: Would you like details on any specific integration?"

Tips for Effective Group Chats

Name agents descriptively - Agent names appear in the chat, so "Frontend-React" is clearer than "Agent1"
Be specific in questions - The more context you provide, the better the moderator can route
@mention explicitly - You can direct questions to specific agents: "What does @Backend think?"
Let the moderator work - It may take multiple rounds for complex questions

Achievements

Maestro features a conductor-themed achievement system that tracks your cumulative Auto Run time. The focus is simple: longest run wins. As you accumulate Auto Run hours, you level up through 11 ranks inspired by the hierarchy of orchestral conductors.

Conductor Ranks

Level	Rank	Time Required	Example Conductor
1	Apprentice Conductor	15 minutes	Gustavo Dudamel (early career)
2	Assistant Conductor	1 hour	Marin Alsop
3	Associate Conductor	8 hours	Yannick Nézet-Séguin
4	Resident Conductor	24 hours	Jaap van Zweden
5	Principal Guest Conductor	1 week	Esa-Pekka Salonen
6	Chief Conductor	30 days	Andris Nelsons
7	Music Director	3 months	Sir Simon Rattle
8	Maestro Emeritus	6 months	Bernard Haitink
9	World Maestro	1 year	Kirill Petrenko
10	Grand Maestro	5 years	Riccardo Muti
11	Titan of the Baton	10 years	Leonard Bernstein

Reaching the Top

Since Auto Runs can execute in parallel across multiple Maestro sessions, achieving Titan of the Baton (Level 11) is technically feasible in less than 10 calendar years. Run 10 agents simultaneously with worktrees and you could theoretically hit that milestone in about a year of real time.

But let’s be real—getting to Level 11 is going to take some serious hacking. You’ll need a well-orchestrated fleet of agents running around the clock, carefully crafted playbooks that loop indefinitely, and the infrastructure to keep it all humming. It’s the ultimate test of your Maestro skills.

The achievement panel shows your current rank, progress to the next level, and total accumulated time. Each rank includes flavor text and information about a legendary conductor who exemplifies that level of mastery.

Command Line Interface

Maestro includes a CLI tool (maestro-cli) for managing agents and running playbooks from the command line, cron jobs, or CI/CD pipelines. The CLI requires Node.js (which you already have if you’re using Claude Code).

Installation

The CLI is bundled with Maestro as a JavaScript file. Create a shell wrapper to run it:

# macOS (after installing Maestro.app)
echo '#!/bin/bash\nnode "/Applications/Maestro.app/Contents/Resources/maestro-cli.js" "$@"' | sudo tee /usr/local/bin/maestro-cli && sudo chmod +x /usr/local/bin/maestro-cli

# Linux (deb/rpm installs to /opt)
echo '#!/bin/bash\nnode "/opt/Maestro/resources/maestro-cli.js" "$@"' | sudo tee /usr/local/bin/maestro-cli && sudo chmod +x /usr/local/bin/maestro-cli

# Windows (PowerShell as Administrator) - create a batch file
@"
@echo off
node "%ProgramFiles%\Maestro\resources\maestro-cli.js" %*
"@ | Out-File -FilePath "$env:ProgramFiles\Maestro\maestro-cli.cmd" -Encoding ASCII

Alternatively, run directly with Node.js:

node "/Applications/Maestro.app/Contents/Resources/maestro-cli.js" list groups

Usage

# List all groups
maestro-cli list groups

# List all agents
maestro-cli list agents
maestro-cli list agents --group <group-id>

# Show agent details (history, usage stats, cost)
maestro-cli show agent <agent-id>

# List all playbooks (or filter by agent)
maestro-cli list playbooks
maestro-cli list playbooks --agent <agent-id>

# Show playbook details
maestro-cli show playbook <playbook-id>

# Run a playbook
maestro-cli playbook <playbook-id>

# Dry run (shows what would be executed)
maestro-cli playbook <playbook-id> --dry-run

# Run without writing to history
maestro-cli playbook <playbook-id> --no-history

# Wait for agent if busy, with verbose output
maestro-cli playbook <playbook-id> --wait --verbose

# Debug mode for troubleshooting
maestro-cli playbook <playbook-id> --debug

JSON Output

By default, commands output human-readable formatted text. Use --json for machine-parseable JSONL output:

# Human-readable output (default)
maestro-cli list groups
GROUPS (2)

🎨  Frontend
group-abc123
⚙️  Backend
group-def456

# JSON output for scripting
maestro-cli list groups --json
{"type":"group","id":"group-abc123","name":"Frontend","emoji":"🎨","timestamp":...}
{"type":"group","id":"group-def456","name":"Backend","emoji":"⚙️","timestamp":...}

# Running a playbook with JSON streams events
maestro-cli playbook <playbook-id> --json
{"type":"start","timestamp":...,"playbook":{...}}
{"type":"document_start","timestamp":...,"document":"tasks.md","taskCount":5}
{"type":"task_start","timestamp":...,"taskIndex":0}
{"type":"task_complete","timestamp":...,"success":true,"summary":"...","elapsedMs":8000}
{"type":"document_complete","timestamp":...,"tasksCompleted":5}
{"type":"complete","timestamp":...,"totalTasksCompleted":5,"totalElapsedMs":60000}

Scheduling with Cron

# Run a playbook every hour (use --json for log parsing)
0 * * * * /usr/local/bin/maestro-cli playbook <playbook-id> --json >> /var/log/maestro.jsonl 2>&1

Requirements

At least one AI agent CLI must be installed and in PATH (Claude Code, Codex, or OpenCode)
Maestro config files must exist (created automatically when you use the GUI)

Provider Nuances

Each AI agent has unique capabilities and limitations. Maestro adapts its UI based on what each provider supports.

Claude Code

Feature	Support
Image attachments	✅ New and resumed sessions
Session resume	✅ `--resume` flag
Read-only mode	✅ `--permission-mode plan`
Slash commands	✅ `/help`, `/compact`, etc.
Cost tracking	✅ Full cost breakdown
Model selection	✅ `--model` flag (via custom CLI args)

OpenAI Codex

Feature	Support
Image attachments	⚠️ New sessions only (not on resume)
Session resume	✅ `exec resume <id>`
Read-only mode	✅ `--sandbox read-only`
Slash commands	⚠️ Interactive TUI only (not in exec mode)
Cost tracking	❌ Token counts only (no pricing)
Model selection	✅ `-m, --model` flag

Notes:

Codex’s resume subcommand doesn’t accept the -i/--image flag. Images can only be attached when starting a new session. Maestro hides the attach image button when resuming Codex sessions.
Codex has slash commands (/compact, /undo, /diff, etc.) but they only work in interactive TUI mode, not in exec mode which Maestro uses.

OpenCode

Feature	Support
Image attachments	✅ New and resumed sessions
Session resume	✅ `--session` flag
Read-only mode	✅ `--agent plan`
Slash commands	❌ Not investigated
Cost tracking	✅ Per-step costs
Model selection	✅ `--model provider/model`

Note: OpenCode uses the run subcommand which auto-approves all permissions (similar to Codex’s YOLO mode).

Configuration

Settings are stored in:

macOS: ~/Library/Application Support/maestro/
Windows: %APPDATA%/maestro/
Linux: ~/.config/maestro/

Cross-Device Sync (Beta)

Maestro can sync settings, sessions, and groups across multiple devices by storing them in a cloud-synced folder like iCloud Drive, Dropbox, or OneDrive.

Setup:

Open Settings (Cmd+,) → General tab
Scroll to Storage Location
Click Choose Folder... and select a synced folder:

iCloud Drive: ~/Library/Mobile Documents/com~apple~CloudDocs/Maestro
Dropbox: ~/Dropbox/Maestro
OneDrive: ~/OneDrive/Maestro

Maestro will migrate your existing settings to the new location
Restart Maestro for changes to take effect
Repeat on your other devices, selecting the same synced folder

What syncs:

Settings and preferences
Session configurations
Groups and organization
Agent configurations
Session origins and metadata

What stays local:

Window size and position (device-specific)
The bootstrap file that points to your sync location

Important limitations:

Single-device usage: Only run Maestro on one device at a time. Running simultaneously on multiple devices can cause sync conflicts where the last write wins.
No conflict resolution: If settings are modified on two devices before syncing completes, one set of changes will be lost.
Restart required: Changes to storage location require an app restart to take effect.

To reset to the default location, click Use Default in the Storage Location settings.

Remote Access

Maestro includes a built-in web server for mobile remote control:

Automatic Security: Web server runs on a random port with an auto-generated security token embedded in the URL
QR Code Access: Scan a QR code to connect instantly from your phone
Global Access: All sessions are accessible when the web interface is enabled - the security token protects access
Remote Tunneling: Access Maestro from anywhere via Cloudflare tunnel (requires cloudflared CLI)

Mobile Web Interface

The mobile web interface provides:

Real-time session monitoring and command input
Device color scheme preference support (light/dark mode)
Connection status indicator with automatic reconnection
Offline queue for commands typed while disconnected
Swipe gestures for common actions
Quick actions menu for the send button

Local Access (Same Network)

Click the "OFFLINE" button in the header to enable the web interface
The button changes to "LIVE" and shows a QR code overlay
Scan the QR code or copy the secure URL to access from your phone on the same network

Remote Access (Outside Your Network)

To access Maestro from outside your local network (e.g., on mobile data or from another location):

Install cloudflared: brew install cloudflared (macOS) or download for other platforms
Enable the web interface (OFFLINE → LIVE)
Toggle "Remote Access" in the Live overlay
A secure Cloudflare tunnel URL will be generated
Use the Local/Remote pill selector to switch between QR codes
The tunnel stays active as long as Maestro is running - no time limits, no account required

Troubleshooting & Support

Debug Package

If you encounter deep-seated issues that are difficult to diagnose, Maestro can generate a Debug Package—a compressed bundle of diagnostic information that you can safely share when reporting bugs.

To create a Debug Package:

Press Cmd+K (Mac) or Ctrl+K (Windows/Linux) to open Quick Actions
Search for "Create Debug Package"
Choose a save location for the .zip file
Attach the file to your GitHub issue

What’s Included

The debug package collects metadata and configuration—never your conversations or sensitive data:

File	Contents
`system-info.json`	OS, CPU, memory, Electron/Node versions, app uptime
`settings.json`	App preferences with sensitive values redacted
`agents.json`	Agent configurations, availability, and capability flags
`external-tools.json`	Shell, git, GitHub CLI, and cloudflared availability
`sessions.json`	Session metadata (names, states, tab counts—no conversations)
`processes.json`	Active process information
`logs.json`	Recent system log entries
`errors.json`	Current error states and recent error events
`storage-info.json`	Storage paths and sizes

Privacy Protections

The debug package is designed to be safe to share publicly:

API keys and tokens — Replaced with [REDACTED]
Passwords and secrets — Never included
Conversation content — Excluded entirely (no AI responses, no user messages)
File contents — Not included from your projects
Custom prompts — Not included (may contain sensitive context)
File paths — Sanitized to replace your username with ~
Environment variables — Only counts shown, not values (may contain secrets)
Custom agent arguments — Only [SET] or [NOT SET] shown, not actual values

Example path sanitization:

Before: /Users/johndoe/Projects/MyApp
After: ~/Projects/MyApp

Getting Help

GitHub Issues: Report bugs or request features
Discord: Join the community
Documentation: This README, CONTRIBUTING.md, and ARCHITECTURE.md

Contributing

See CONTRIBUTING.md for development setup, architecture details, and contribution guidelines.

Contributors

@pedramamini - Creator and maintainer
@mattjay - Contributor and tester

License

AGPL-3.0 License