Is my code encrypted before it leaves my machine?

Yes. The Privatemode proxy runs locally and encrypts all code context, prompts, and responses before they leave your machine. It also performs remote attestation to verify the integrity of the Privatemode service, ensuring end-to-end protection for every AI request Zed makes.

Does Privatemode store or learn from my code?

No. Privatemode never learns from your data. All requests are processed using confidential computing, and no code, prompts, or responses are retained after processing. This goes beyond policy promises — the confidential computing architecture enforces it, and reproducible builds make it verifiable.

Do I still need the Privatemode proxy with Zed?

Yes. The proxy is a lightweight local process that handles encryption and remote attestation. It verifies that the Privatemode service is running in a genuine confidential computing environment before sending any data. Zed connects to the proxy as if it were a standard OpenAI endpoint.

How does this compare to using Zed’s built-in AI providers?

Zed’s built-in providers (Anthropic, OpenAI, etc.) send your code context directly to those services. While Zed states they don’t train on your data, the providers still process your code in plaintext. Privatemode adds technical protection: your code is encrypted end-to-end and processed inside a hardware-isolated environment whose memory stays encrypted, so by design, neither the infrastructure provider nor Privatemode can read your data.

Integrating Privatemode AI with Zed

Zed's Agent Panel and Inline Assistant both reason over your code. Pointing them at Privatemode keeps those features and swaps the model for open-weight ones running in a hardware-isolated environment whose memory stays encrypted.

Introduction

The challenge

Zed's AI features send your code to a provider

Zed's Agent Panel and Inline Assistant read your code and send it to the configured provider, and with a conventional one it is processed in plaintext on that provider's servers. Some developers have moved to forks such as Zedless that remove the AI features entirely, trading capability for control.

The solution

Keep the features, change the provider

Zed lets you add an OpenAI-compatible provider in its settings, pointed at the Privatemode proxy on your machine. The models behind it support tool calling, so the Agent Panel keeps working as before. The proxy verifies the service through remote attestation and encrypts each request before it leaves.

Benefits

Why use Privatemode AI with Zed?

Agent Panel keeps working

See models

End-to-end confidentiality

Learn how

Free tier available

How to get started

How to set up Privatemode in Zed

Get your API key

If you don’t have a Privatemode API key yet, you can generate one for free here.

Run the Privatemode proxy

The proxy verifies the integrity of the Privatemode service using remote attestation. The proxy also encrypts all data before sending and decrypts data it receives.

Configure Zed to use Privatemode

Open Zed and go to Settings (Cmd+,). Zed natively supports OpenAI-compatible endpoints, so add Privatemode as a custom provider in your settings.json under language_models.openai_compatible, with api_url pointing at the proxy's /v1 endpoint. Then enter any placeholder as the API key in the Agent Panel settings (command agent: open settings) or set the PRIVATEMODE_API_KEY environment variable: the proxy is already holding your real key.

capabilities.tools is what lets the Agent Panel edit files and run commands rather than only chat.

Walk through our interactive demo

FAQ

Frequently asked questions about using Privatemode with Zed

The Agent Panel and the Inline Assistant. Privatemode is added as an OpenAI-compatible provider, which Zed supports natively, so both use it once it is configured in settings.json. Edit Prediction is configured separately under Zed's edit_predictions setting and is not covered by this setup.