Maestri
Maestri is a macOS terminal app that orchestrates multiple AI agents, terminals, sketches, and notes on one infinite canvas—reduce context switching.
What is Maestri?
Maestri is a macOS terminal app designed to orchestrate multiple AI agents and related artifacts (terminals, sketches, and notes) on a single “infinite canvas.” Its core purpose is to reduce the context switching that happens when working with several agent-driven terminal sessions at once.
Instead of managing separate terminal tabs and scattered notes, Maestri places each agent and its associated terminals in a shared spatial workspace. Agents can also connect to each other, with on-device help from “Ombro” that summarizes completed work and suggests next steps.
Key Features
- Infinite canvas workspace for agent workflows: pan, zoom, focus, and spatially organize related work so you can keep context across multiple agents.
- Multi-agent orchestration on one canvas: spawn agents across the canvas, with each one running its own terminal/agent (including options mentioned such as Claude Code, Codex, OpenCode, or a plain shell).
- Terminal-to-note integration: connect a terminal to a sticky note so an agent can write directly to it, capturing updates, summaries, and decisions while you continue working.
- Agent connections and delegation: connect terminals so agents can prompt each other directly, including handoffs, questions, and task delegation.
- Reusable agent roles with custom instructions: define roles (e.g., a lead, coder, reviewer, tester) and assign them to agents so each has a defined responsibility.
- Sketching tools for architecture and flow: draw diagrams and arrows directly on the canvas near the agents doing the work.
- Cross-app navigation to edit code: one-click open to jump from an agent’s terminal work into VS Code, Zed, Xcode, or another preferred editor.
- Workspaces and tmux-style shortcuts: separate canvases per project and use shortcuts to jump between workspaces that remember where you left off.
- Ombr0 on-device companion for agent status: Ombro monitors agents and shows summaries and next-step suggestions in a floating window outside the app.
How to Use Maestri
- Download Maestri for macOS (requires macOS 26.2+ and Apple Silicon per the site).
- Start with a workspace and add agents onto the canvas; you can place terminals anywhere in the canvas so related work stays near each other.
- Connect terminals to sticky notes when you want agent-written progress, summaries, or decisions captured alongside the work.
- Use agent roles (custom instructions) to assign responsibilities to each agent, and connect agents when you want them to coordinate prompts with less manual intervention.
- When you’re ready to edit, use the one-click action to open the project in your editor (VS Code, Zed, Xcode, or your preferred editor).
Use Cases
- Coordinating multiple coding agents for one feature: run separate agents for coding, review, and testing on the same canvas, with notes placed next to the relevant terminal sessions.
- Architecture planning with working context: sketch component diagrams and data-flow arrows on the canvas while the associated agents execute tasks, keeping planning and implementation visible together.
- Reducing tab switching across projects: keep separate canvases for separate projects and jump between them using tmux-style shortcuts that preserve your last position.
- Capturing agent outputs as you work: connect a terminal session to a sticky note so summaries and decisions are written directly into notes while you continue other tasks.
- Guided follow-up on completed tasks: rely on Ombro to summarize what an agent finished and suggest what to do next in a floating window so you can continue working in other apps.
FAQ
-
Does Maestri require an account? No account is required to download and start using it.
-
Is Maestri’s AI processing on-device? Ombro runs on-device using Apple Foundation Models, and the site states it runs entirely on the Mac with no API calls and no cloud.
-
Is there telemetry or analytics? The site states there is zero telemetry: no analytics, no tracking, and no usage data collected.
-
What systems does Maestri support? The site lists macOS 26.2+ and Apple Silicon.
-
Can agents interact with each other? Yes. The app supports “agent connections,” where terminals can be connected so agents can prompt each other directly.
Alternatives
- Standard terminal multiplexers (e.g., tmux/screen) plus an editor and notes app: useful for managing multiple sessions, but typically rely on manual switching and external note organization rather than a unified spatial canvas.
- IDE-integrated AI assistance or agent features: can provide in-editor help and code actions, but may not offer the same canvas-based coordination of multiple agents, sketches, and notes together.
- General-purpose knowledge/workspace tools with AI notes: good for storing and visualizing ideas, but may not tightly integrate with terminals and multi-agent execution the way Maestri does.
- Other AI agent orchestration tools that run separate sessions: can coordinate agents, but may still keep outputs in separate tabs/windows instead of one shared pan-and-zoom canvas.
Alternatives
Biji
Biji is a versatile platform designed to enhance productivity through innovative tools and features.
Tavus
Tavus builds AI systems for real-time, face-to-face interactions that can see, hear, and respond, with APIs for video agents, twins & companions.
Falconer
Falconer is a self-updating knowledge platform for high-speed teams to write, share, and find reliable internal documentation and code context in one place.
HiringPartner.ai
HiringPartner.ai is an autonomous recruiting platform with AI agents that source, screen, call, and interview candidates 24/7, reducing time-to-hire from weeks to as little as 48 hours.
OpenFlags
OpenFlags is an open source, self-hosted feature flag system with a control plane and typed SDKs for progressive delivery and safe rollouts.
Decision Jar
Create virtual jars of options with Decision Jar—shake your phone to pick instantly, use AI suggestions, and view decision logs over time.