Tutti — Real-time shared workspace for people and agents
A real-time shared workspace for your AI agents, where context, files, apps, and tasks stay connected—so agents build in tune, not in silos.
- Source repo
- tutti-os/tutti
- Stars
- ★ 3.8k
- Last updated
- 11d ago
- License
- Apache-2.0
- Primary language
- TypeScript
- FA score
- 31/100 · Major gaps
At a glance
- How it runs
- Works with
- Portable with changesCodex · Claude CodeChatGPT (Unsupported)
- Cost
- Free software; you pay for model usage
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- Independent developers can let Claude draft a plan and Codex carry development forward, avoiding re-explaining project context.
- Not a fit if
- Teams relying only on unsupported agents like ChatGPT
- Teams needing Windows support or the released cloud Room
- Source review
- 31/100 · Major gaps 8 safety controls not found
What does this agent do, and when should you use it?
Tutti is an open-source desktop application (currently macOS only) that solves context handoff between multiple AI coding agents like Claude Code, Codex, and Hermes. It provides a real-time shared workspace where agent conversations, files, tasks, and app outputs are all connected. With @ and + references, agents can pick up each other's work without losing context, eliminating manual copy-pasting and re-explanation. Tutti includes an app center with image generation, UI/UX design, document, and presentation apps that both users and agents can call, with outputs flowing directly into the workspace. It runs on your existing agent subscriptions, avoiding extra costs. The upcoming Tutti · VM adds multi-user, multi-device collaboration through a cloud-based shared Room. Licensed under Apache-2.0.
Tutti orchestrates a real-time shared workspace that connects multiple agents, files, running tasks, and apps. It enables @ references, allowing one agent (e.g., Codex) to reference conversations, files, app invocations, and tasks from another agent (e.g., Claude Code) and even command that agent to act. The + reference lets agents attach local files or app-generated outputs directly into the chat. A Control Center gives a unified view of all agent conversations, pending approvals, and running tasks. Tutti features a built-in app ecosystem including AI Canvas (image generation) and AI PPT, which both users and agents can invoke; outputs remain in the workspace for subsequent steps. It also supports goal-to-task decomposition: describing a goal generates sub-tasks that can be assigned to suitable agents. Tutti leverages existing subscriptions for Claude Code, Codex, and Hermes (OpenClaw in development) without additional cost.
- Independent developers can let Claude draft a plan and Codex carry development forward, avoiding re-explaining project context.
- Designers can create mockups in a design app and have Codex convert them into working code directly.
- Product managers can have Codex write a PRD and then call a UI/UX design app to generate a prototype, no Figma needed.
- Content creators can produce scripts, visuals, and layouts end-to-end in one shared workspace.
- Teams using multiple agents (Codex, Claude Code) can work on the same project with shared context, reducing coordination overhead.
How do you install or deploy this agent?
Download and install the macOS app from https://tutti.sh/desktop/download?platform=macos&arch=universal&format=dmg. Windows support is coming soon. No command-line installer is currently provided.
How do you use this agent?
Launch Tutti and connect your existing agent subscriptions (e.g., Claude Code, Codex). In any agent chat box, use @ to reference other conversations, files, tasks, or app invocations; use + to attach local files or app outputs. Monitor all activity in the Control Center. Use the goal-to-task feature to break a high-level goal into sub-tasks and assign them to appropriate agents. Invoke built-in apps like AI Canvas or AI PPT, and reference their outputs in subsequent steps.
What are this agent's strengths and limitations?
- Real-time shared workspace eliminates context loss, enabling seamless handoffs between agents.
- @ and + references reduce copy-pasting, making multi-agent workflows significantly more efficient.
- Built-in app center (image, design, docs, PPT) adds capabilities without additional subscriptions.
- Local-first architecture keeps data on your machine, enhancing privacy and security.
- Only macOS is supported; Windows version is not yet available.
- Limited to specific agents (Claude Code, Codex, Hermes); no support for ChatGPT or other platforms.
- Requires existing subscriptions for these agents; without them, you may incur costs via Tutti Agent.
- Cloud collaboration features (Tutti · VM) are not yet released, limiting cross-device/multi-user scenarios.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Tutti — Real-time shared workspace for people and agents This agent | 31 · Major gaps | Desktop appFree + model costs | ★ 3.8k | 11d ago | TypeScript | Codex · Claude Code |
| Solo Agent | 48 · Major gaps | Self-hosted serviceFree + model costs | ★ 591 | 1mo ago | Go | Codex · Claude Code |
| Agentlas OS | 46 · Major gaps | CLIFree + model costs | ★ 1.6k | 2d ago | Python | Codex · Claude Code |
| Munder Difflin – Multi-Agent Office Floor | 43 · Major gaps | Desktop appFree + model costs | ★ 8.9k | today | TypeScript | ChatGPT · Codex · Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows the repository includes a security policy (SECURITY.md) and release workflows, but no detailed description of permission model, user confirmation mechanisms, data flow transparency, sensitive data handling, dependency security audits, external effect controls, rollback mechanisms, or source attribution. Therefore, all trust criteria score 0.
Self-consistency: README and package.json are generally consistent in description, but no detailed architecture documentation or behavioral specifications are provided, so score 1. Dependency availability: package.json lists dependencies and pnpm overrides, but no availability guarantees or mirrors are provided, score 1. Failure messages: SECURITY.md and test files provide error handling examples, but no comprehensive failure message strategy, score 1.
Audience and scenarios: README describes target users and use cases in detail, score 2. Capability boundaries: README states Tutti is not a replacement for coding agents, but does not clearly define capability boundaries, score 1. Trigger precision: No detailed description of trigger mechanisms or commands, score 1. Environment fit: README mentions macOS support and Windows coming soon, but no detailed environment requirements, score 1.
Information architecture: README and docs directory provide clear navigation, score 2. Install notes: README provides download links but no detailed installation steps, score 1. Naming stability: No naming conventions or version naming strategy provided, score 1. Examples and FAQ: README includes FAQ and examples, score 2. Known limitations: README mentions Windows support coming soon but does not list other limitations, score 1. License: LICENSE file is Apache-2.0, score 3. Versioning and changelog: No CHANGELOG or version control instructions provided, score 1. Maintenance responsibility: SECURITY.md and CONTRIBUTING.md provide maintenance guidelines but do not clearly define maintenance responsibility, score 1.
Output usability: README describes how outputs can be referenced but does not provide specific output formats, score 1. Marginal value: README emphasizes the value of reducing context switching, score 2. Cost benefit: README mentions using existing subscriptions but does not provide cost analysis, score 1.
Claim traceability: Claims in README are not supported by specific evidence or links, score 1. Cross-source corroboration: No third-party verification or independent testing provided, score 1. Fact-inference separation: README mixes facts and inferences without clear distinction, score 1.
- Not found in source: least-privilege scopingGrant only what the task needs: a dedicated account or read-only token, scoped to specific directories and repos.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Not found in source: data-flow disclosureWatch which external services it contacts (proxy or firewall logs) and keep sensitive data out until you know where it goes.
- Not found in source: sensitive-data handlingUse dedicated, low-privilege, revocable API keys — never production credentials — and keep secrets out of logs.
- Not found in source: dependency securityPin versions and run a dependency audit (npm audit, pip-audit) before installing; prefer running it in a container.
- Not found in source: disclosed external effectsEstablish which external systems it writes to, sends to or changes, and verify with test accounts or repos before production.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Not found in source: verifiable attributionInstall from the official repo or registry and check the publisher and URL to avoid look-alike packages.
- The repository does not provide a permission model or user confirmation mechanisms, which may not satisfy the principle of least privilege.
- Dependency security audits and rollback mechanisms are not documented, requiring further review.
- Release workflows involve sensitive credentials, but no detailed key management instructions are provided.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
stable Stable (Recommended)
Release notes excerpt
<!-- tutti-desktop-stable-release-alias:start --> Stable Desktop Release Current stable release: [v0.2.33](https://github.com/tutti-os/tutti/releases/tag/v0.2.33) This floating release is refreshed after every desktop release. GitHub Releases is reserved for the recommended stable build; RC and beta downloads are distributed through their preview channels. <!-- tutti-desktop-stable-release-alias:end --> <!-- tutti-desktop-release-review:start --> Release Notes 中文 v0.2.33 稳定版发布 问题修复 - 修复了 Agent 与 Codex CLI 版本不兼容的问题，确保使用当前支持的 Codex CLI 版本 发布与更新 …View on GitHub ↗ -
Release notes excerpt
<!-- tutti-desktop-release-review:start --> Release Notes 中文 v0.2.33 稳定版发布 问题修复 - 修复了 Agent 与 Codex CLI 版本不兼容的问题，确保使用当前支持的 Codex CLI 版本 发布与更新 - 改进了 macOS 打包流程，将构建环境固定到稳定 runner，提升打包可靠性 English v0.2.33 Stable Release Bug Fixes - Fixed Agent compatibility with Codex CLI by requiring the current supported Codex CLI version Release and Updates - Improved macOS packaging by pinning the build environment to a stable runner for more reliable builds <!-- tutti-desktop-release-review:end --> What's Changed …View on GitHub ↗ -
Release notes excerpt
<!-- tutti-desktop-release-review:start --> Release Notes 中文 v0.2.32 稳定版发布 问题修复 - 修复了外部 Claude 命令丢失的问题，自定义命令现在可以正常保留 - 改进了 Claude 二进制文件的安装与状态检测流程，Agent 启动更可靠 体验优化 - 优化了 Windows 平台上的 Agent 运行环境准备流程 - 完善了 Agent 提供商配置的相关指引 发布与更新 - 内部改进了 Claude 代码锁定与二进制打包流程，提升更新验证可靠性 English v0.2.32 Stable Release Bug Fixes - Fixed external Claude commands being lost; custom commands are now preserved correctly - Improved Claude binary installation and status detection for more reliable Agent startup Experience Improvements …View on GitHub ↗