Zed v1.8 Ships a Bootstrap Hook That Automates AI Agent Environments

Zed v1.8 is out, and the headline feature is deceptively small: a single new setting called agent.terminal_init_command that runs a shell command every time the agent opens a fresh terminal thread. It sounds minor, but it unlocks a whole class of workflows that were awkward before -- environment setup, virtual-env activation, toolchain switching -- all handled automatically before the agent types its first command.

The agent terminal gets a bootstrap hook

When Zed's AI agent runs tasks in a terminal, it previously started from a bare shell. That meant any project-specific environment setup (activating a Python venv, loading a .env file, switching Node versions) had to happen manually or be baked into every prompt. With agent.terminal_init_command, you configure it once in your settings and every new agent terminal thread inherits the right environment automatically.

To use it, add this to your Zed settings:

{
  "agent": {
    "terminal_init_command": "source .venv/bin/activate && export ENV=dev"
  }
}

This is particularly useful for Python projects, monorepos with multiple toolchains, or any setup where the agent needs specific environment variables to run commands correctly.

Git gets noticeably faster

The bulk of the engineering effort in this release went into Git performance. The team tackled several specific bottlenecks:

• Faster HEAD resolution -- the lookup that happens constantly as you work
• Faster uncommit, remote listing, and default-branch lookups
• Fewer git processes spawned overall as you edit
• Improved scrolling performance in diff views containing large hunks with many computed word diffs, and improved performance of the project diff especially in the presence of merge conflicts

If you work in large repos or repos with complex histories, this release should feel meaningfully snappier in the Git panel and diff views. The team also fixed a long-standing issue where the number of git operations triggered when repository state changes outside of Zed was reduced , which matters for teams using external tools like GitHub Desktop or CLI workflows alongside the editor.

Two new selection commands worth knowing

The new editor: select inside delimiters and editor: select around delimiters actions select content within the nearest enclosing brackets, braces, parentheses, or quotes -- and repeating the action expands the selection to the next enclosing pair. This is the Zed equivalent of Vim's ci( / ca( motions, now available as first-class editor commands bindable to any key.

The other new command, workspace: reset pane sizes, equalizes the sizes of all panes in the center group -- a small quality-of-life fix for anyone who has accidentally dragged a pane divider into an awkward layout.

Editor polish across the board

Beyond the headliners, v1.8 includes a round of polish that adds up:

• Improved markdown preview styles, making it more legible and easier to parse
• Improved Zed's memory usage when a language server emits messages faster than the foreground thread can handle them, and improved editor: go to definition to avoid repeatedly querying the language server while moving the mouse over the same symbol
• Diff scrolling, multi-cursor editing, and markdown preview search are all snappier
• Prettier now reads the project's .editorconfig files when formatting

Availability and pricing

The stable release is available now for macOS, Windows, and Linux via zed.dev/download. The core editor is free forever. Pro at $10 per month unlocks unlimited edit predictions, $5 of included tokens for Zed-hosted models, and usage-based billing beyond that at API list price plus 10 percent. A two-week free trial of the Pro plan comes with $20 of token credits and unlimited accepted edit predictions.

For teams that prefer to bring their own keys, you can bring your own LLM API key for unlimited AI use on the free plan. The v1.8 update is delivered automatically for existing installs or installable fresh via the download page.