Zed 1.0 editor showing the new interface with agent threads and parallel sessions Image: Zed Industries
by VibecodedThis

Zed Reaches 1.0 and Introduces an Open Protocol for Connecting AI Agents to Editors

Zed shipped version 1.0 on April 29, reaching the milestone with cross-platform support, a new Agent Client Protocol backed by Google and JetBrains, and a Zed for Business tier.

Share

Zed hit version 1.0 on April 29, five years and more than a million lines of Rust after the editor’s founding. The milestone release ships on macOS, Windows, and Linux with full cross-platform support, parallel agent execution, and something the team is treating as more significant in the long run: the Agent Client Protocol.

Nathan Sobo, Zed’s CEO, put it plainly in the release post: “1.0 doesn’t mean ‘done’. It also doesn’t mean ‘perfect’. It means we’ve reached a tipping point where most developers can quickly feel at home in Zed.”

The Agent Client Protocol

The Agent Client Protocol (ACP) is an open standard for connecting AI agents to code editors. Zed built the protocol and published it under the Apache license, then brought in Google and JetBrains as partners to help develop and standardize it across the industry.

The basic idea: any agent that implements ACP gains access to multi-file editing, full codebase context, and code review tooling within compatible editors, without being tied to any single vendor’s plugin system. The protocol is designed so that “nothing touches our servers” when using a third-party agent, meaning Zed doesn’t proxy or inspect the agent communication.

The current list of supported agents spans more than 30 tools, including Claude Agent (Anthropic), Gemini CLI (Google), GitHub Copilot, Codex, OpenCode, and Cline. The ACP specification is publicly documented and other editors can implement it.

On the editor side, Zed is the primary implementation. JetBrains integration is in progress, and the protocol is also being picked up by Neovim plugins, Emacs, and a handful of newer tools. Gemini CLI is listed as the reference implementation.

External Agents in Practice

Before 1.0, Zed’s AI features were mostly internal: the editor’s own agents running against models you configured directly. ACP changes the framing. External agents run as separate processes and communicate with Zed via the protocol. From the user’s perspective, you pick an agent from a list and it integrates into the same editing surface Zed’s native agents use.

The external agents documentation lists Claude Agent, Gemini CLI, Codex, and GitHub Copilot as officially supported options. For teams already running one of those agents in their workflow, it means switching editors doesn’t mean losing the agent setup.

Zed for Business

Zed for Business launched alongside 1.0. It adds centralized billing and seat management, role-based access controls for AI models, spend tracking per member, and data policy controls for teams with security requirements.

The business tier is aimed at engineering organizations that want to standardize on Zed and control which models individual contributors can use, with visibility into AI spending across the team.

Other 1.0 Features

The release also includes bookmarks, a git: view commit command palette action, GIF animation in Markdown preview, and support for DeepSeek-V4-Pro and DeepSeek-V4-Flash models.

Parallel agent support shipped earlier, in April, but is part of the 1.0 feature set. Multiple agents can run concurrently in the same window on different parts of the codebase. The Threads Sidebar groups sessions by project with independent access controls per thread.

Performance numbers from the Zed team: 0.4-second startup, 2-millisecond input latency, and 120fps rendering. That’s the baseline they’ve maintained even with agents running, which matters for a feature that’s inherently concurrent.

Context

Zed’s founding pitch was a fast, clean alternative to Electron-based editors. The agentic additions over the past year represent a significant expansion of scope. The Agent Client Protocol, in particular, positions Zed less as a self-contained product and more as an interoperability layer for a broader ecosystem. Whether that resonates with JetBrains and other editor teams will determine how far the protocol actually travels.

Sources:

Share