Documentation — Maestro Studio

User manual

Complete guide to the Maestro desktop IDE, agent platform, service, and remote console.

Install Maestro, learn the desktop layout, create a project, and run a first workflow safely.

Configure models and routes, author typed workflows, use MRGD, and inspect execution history.

Use persistent chat, voice, rooms, memory, skills, automations, tasks, and messaging channels.

Use files, terminals, execution backends, coding runners, web tools, MCP, custom integrations, and trajectory exports.

Manage accounts, profiles, security, service deployment, remote console, runtime health, backups, and signed software updates.

Feature index, shortcuts, types, platform support, data locations, troubleshooting, and glossary.

Create a project, register your first models, store keys safely, and run a three-node workflow — the quarter-hour tour.

A research → draft → review pipeline with two agents, a human approval gate, and a budget that keeps it honest.

Rules, conditions, fallback chains, and cost-aware routing — and how to route image, video, and speech tasks to different models in seconds.

Enable reward-guided decoding on a node, pick scorers, set k and the precision/recall weight, and read the inspector like a pro.

Pair your account with a code, talk to your agent from your phone, and schedule a daily briefing that arrives while the IDE is closed.

Teach agents once: review and approve self-drafted skills, manage what agents remember, and search everything you've ever discussed.

What MRGD is, how the algorithm works, what the paper measured, and how Maestro implements it.

Platforms, availability, security, offline use, budgets, and pricing status — answered plainly.