Cindy is an open-source AI agent, ready to work. Connect your computers and phone, share access to tasks, files and models, and bring your favorite agents and existing subscriptions together.
Your work computer, home machine, laptop and phone, connected. Change screens without losing the task. Reach your conversations and files, and use models configured on another computer.
Use subscriptions, APIs and local models configured on another computer. Keep the task on its original machine and borrow model access from another, without repeating the setup everywhere.
Access the same task across devices. Browse, preview and transfer task files, retrieve work from your computer, or send it material from your phone.
Built-in remote desktop lets you switch between connected computers. See the actual screen, type and operate software when you need to take over.
Your computers store local project files and full execution records. Working remotely means connecting back to your own environment.
Connect from around the world to your tasks, files and models. No backend to deploy or server to rent. Your chosen model service's charges still apply.
Use the same account with the relevant computers online and authorized. Model sharing also requires remote access enabled for that provider; its fees and limits apply. File access is not whole-drive sync. Online models and cloud companions involve data transfer or cloud sync.
Choose Claude Code, Codex or Pi in the same place. Switch between supported harnesses and models with task history, memory and skills at hand. Less copying context. Less starting over.
Take the best from every stack. Get trapped by none.
Choose from multiple model providers and shape your workflow with extensions and skills. Already integrated in Cindy: pick a model and get to work.
Let another agent pick up the work, inside the same task.
Plan first. Build, test and document in parallel. Then bring in an independent review. Choose a suitable harness and model for each role, follow the work and decide what comes next.
Architecture · Work breakdown · Acceptance criteria
Diff review · Edge cases · Pass / Rework
A harness is an execution engine; models are a separate choice. Available combinations depend on the provider, authorization and model capabilities. This example is not a managed-service model list.
A better agent experience should not mean another model subscription. Connect a plan or API key you already have, or choose Cindy's managed service.
We encourage you to use your own subscriptions and keys.Sign in with Claude, ChatGPT, SuperGrok, Kimi Code, MiniMax or GitHub Copilot. Connect other supported Coding Plans with a key. Pi is a harness, not another subscription to buy.
Your provider's fees and limits still apply. Local models are in beta and depend on your hardware. Third-party tools may require separate authorization.
Work in your existing environment, with files and results close at hand.
Install the connections you need and authorize your accounts.
Pick up tasks on mobile or connect a server environment.
Install useful capabilities and keep your own working methods.
Keep useful context. Teach a method. Add a new capability. Start with an agent ready to work and shape a companion that fits the way you work.
Open source means more than visible. It means changeable.Apache-2.0 · Audit · Fork · Build · Contribute
Keep the roles you work with often. Give each companion a name, a responsibility, memory and skills. Pick up together next time.
Research, implementation or review, with a clear responsibility and working method. No need to rebuild a role prompt every time.
Keep preferences, project context and useful methods in memory and skills you can inspect and improve.
Set up routines on your computer for recurring checks, work and reports. Connected chat channels can bring her new tasks too.
Install skills from SkillHub, or write and publish your own methods for the next task to reuse.
The plugin marketplace is open. Install a connection, inspect its permissions and authorize your account.
Connect internal tools and self-hosted services so Cindy can work in your environment.
Apache-2.0. Suggest changes, contribute code and share skills or plugins so your improvements can help someone else.
Ready out of the box. Never boxed in.
See the process and know the boundaries. Choose permissions, inspect results and decide what comes next.
Set access for each task. Requests that need confirmation wait for you, including on your phone.
View diffs and request an independent review. Eligible local Git tasks can use worktrees and checkpoints.
Files can be handled locally. Model requests send relevant context; companion sync, remote access and external tools also transfer data through their services.
Follow tool calls, results and available usage information. Add direction or stop a task when you need to.
Advanced teams already work alongside AI
Traditional collaboration is a pipeline: product waits for design, design waits for engineering, engineering waits for QA. Give everyone Cindy and put her in the team chat: ideas become discussable drafts on the spot, discussions become active tasks, and the pipeline becomes a network that starts together.
Turn an idea into a clickable demo during the meeting and discuss the artifact, not the spec.
Apply a direction to the real page so the team sees the result, not another mockup.
Run tests, docs and release notes in parallel, then focus on review and judgment.
@ Cindy in Slack or Feishu and the first working version can be there before the meeting ends.
TEAM & ENTERPRISE · JOIN THE WAITLIST
Bring a useful personal workflow to the team. Set up models, permissions, credits and skills centrally so each person does less repeated configuration.
Use your own key or a supported subscription without an extra software fee. Choose Cindy's usage credits or monthly plans when you want a managed model service.
Software features and model usage are separate. Use your own provider and pay them directly, or buy usage credits from Cindy. Managed plans are not unlimited model access.
See full pricing and rules →Cindy brings together the strengths of different harnesses, shaped by our own product vision and attention to detail: intuitive interactions, a cohesive experience across devices, and fluid remote access. Keep your familiar subscriptions and work across your phone and multiple computers.
Cindy is open source and grows with contributions from the community. You can contribute improvements, or shape her through memory, skills, plugins and source code to make Cindy your own.
Claude Code, Codex and Pi are agent harnesses that organize execution. Models do the reasoning and generation. Cindy offers both choices; available combinations depend on the provider, authorization and model capabilities.
Remote tasks execute on your computer, which needs to be online with remote access enabled. Mobile can interact with reachable computers; it does not automatically move offline work into the cloud.
The client is free. Authorize a supported subscription or connect your own API key. Your provider's usage fees and limits still apply. Cindy's managed model service is an optional, separately billed choice.
Use Ollama, llama.cpp or local services such as LM Studio. Local models remain in beta; results depend on the model, hardware and runtime.
Local-first does not mean no data ever leaves your device. Model inference sends necessary conversation and file content to your selected service. Companion cloud sync, remote devices and external plugins also transfer relevant data. Choose services, permissions and sync settings to fit your needs, and read the privacy policy.
Yes. Browse, install and update skills and plugins, or create and share your own. Third-party services may still require their own accounts or subscriptions.
Join the waitlist with your email. We will contact you when access opens. Team focuses on shared setup and governance; Enterprise adds self-hosting and support options.
