Closes#32951
## Summary
When an agent notification was shown and the `AcpThreadView` was dropped
(e.g., by closing the project window or navigating to a new thread), the
notification would become orphaned and undismissable because the
subscriptions handling dismiss events were dropped along with the thread
view.
## Fix
Added an `on_release` callback that closes all notification windows when
the thread view is dropped. This ensures notifications are always
cleaned up properly.
## Testing
Added `test_notification_closed_when_thread_view_dropped` to verify
notifications are closed when the thread view is dropped.
Release Notes:
- Fixed agent notification getting stuck and becoming undismissable when
the project window is closed or when navigating to a new thread
Closes https://github.com/zed-industries/zed/issues/12589
Forces Zed to require user permissions before running any basic
potentially dangerous actions: parsing and synchronizing
`.zed/settings.json`, downloading and spawning any language and MCP
servers (includes `prettier` and `copilot` instances) and all
`NodeRuntime` interactions.
There are more we can add later, among the ideas: DAP downloads on
debugger start, Python virtual environment, etc.
By default, Zed starts in restricted mode and shows a `! Restricted
Mode` in the title bar, no aforementioned actions are executed.
Clicking it or calling `workspace::ToggleWorktreeSecurity` command will
bring a modal to trust worktrees or dismiss the modal:
<img width="1341" height="475" alt="1"
src="https://github.com/user-attachments/assets/4fabe63a-6494-42c7-b0ea-606abb1c0c20"
/>
Agent Panel shows a message too:
<img width="644" height="106" alt="2"
src="https://github.com/user-attachments/assets/0a4554bc-1f1e-455b-b97d-244d7d6a3259"
/>
This works on local, SSH and WSL remote projects, trusted worktrees are
persisted between Zed restarts.
There's a way to clear all persisted trust with
`workspace::ClearTrustedWorktrees`, this will restart Zed.
This mechanism can be turned off with settings:
```jsonc
"session": {
"trust_all_worktrees": true
}
```
in this mode, all worktrees will be trusted by default, allowing all
actions, but no auto trust will be persisted: hence, when the setting is
changed back, auto trusted worktrees will require another trust
confirmation.
This settings switch was added to the onboarding view also.
Release Notes:
- Introduced worktree trust mechanism, can be turned off with
`"session": { "trust_all_worktrees": true }`
---------
Co-authored-by: Matt Miller <mattrx@gmail.com>
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Co-authored-by: John D. Swanson <swanson.john.d@gmail.com>
Fixes#43165
## Problem
MCP prompts were only available in text threads, not agent threads.
Users with MCP servers that expose prompts couldn't use them in the main
agent panel.
## Solution
Added MCP prompt support to agent threads by:
- Creating `ContextServerPromptRegistry` to track MCP prompts from
context servers
- Subscribing to context server events to reload prompts when MCP
servers start/stop
- Converting MCP prompts to available commands that appear in the slash
command menu
- Integrating prompt expansion into the agent message flow
## Testing
Tested with a custom MCP server exposing `explain-code` and
`write-tests` prompts. Prompts now appear in the `/` slash command menu
in agent threads.
Release Notes:
- Added MCP prompt support to agent threads. Prompts from MCP servers
now appear in the slash command menu when typing `/` in agent threads.
---------
Co-authored-by: Agus Zubiaga <agus@zed.dev>
Closes#26823
Release Notes:
- Added support for customising the prompt used for generating commit
message in the rules library
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
This reverts commit f576ec7a15 (#44995).
The optimization introduced a regression that causes the main thread to
hang for **100+ seconds** in certain scenarios, requiring a force quit
to recover.
## Analysis from spindump
When a large `AutoHeight` editor is displayed inside a `List` (e.g.,
Agent Panel thread view), the clipping calculation can produce invalid
row ranges:
1. `visible_bounds` from `window.content_mask().bounds` represents the
window's content mask, not the intersection with the editor
2. When the editor is partially scrolled out of view,
`clipped_top_in_lines` becomes extremely large
3. This causes `start_row` to be computed as an astronomically high
value
4. `blocks_in_range(start_row..end_row)` then spends excessive time in
`Cursor::search_forward` iterating through the block tree
The spindump showed **~46% of samples** (459/1001 over 10+ seconds)
stuck in `BlockSnapshot::blocks_in_range()`, specifically in cursor
iteration.
### Heaviest stack trace
```
EditorElement::prepaint
└─ blocks_in_range + 236
└─ Cursor::search_forward (459 samples)
```
## Symptoms
- Main thread unresponsive for 33-113 seconds before sampling even began
- UI completely frozen
- High CPU usage on main thread (10+ seconds of CPU time in the sample)
- Force quit required to recover
## Path forward
The original optimization goal (reducing line layout work for clipped
editors) is valid, but the implementation needs to:
1. Correctly calculate the **intersection** of editor bounds with the
visible viewport
2. Ensure row calculations stay within valid ranges (clamped to
`max_row`)
3. Handle edge cases where the editor is completely outside the visible
bounds
Release Notes:
- Fixed a hang that could occur when viewing large diffs in the Agent
Panel
This PR adds support for collapsing/expanding Git entries with your
keyboard like you can inside the project panel and variable list.
I noticed there is a bug that selecting the next entry when you are on
the directory level will select a non-visible entry. Will fix that in
another PR, as it is not related to this feature implementation.
**Result**:
https://github.com/user-attachments/assets/912cc146-1e1c-485f-9b60-5ddc0a124696
Release Notes:
- Git panel: Add support for collapsing/expanding entries with your
keyboard.
Until now, Helix-mode users would have to rely on Vim's `d *` behaviour
which cannot be reliably replicated with Helix's default delete
behaviour and so I believe that remapping this functionality to Helix's
goto mode is a better fit.
Release Notes:
- Added custom mappings for Zed specific diff and git-related actions to
Helix's goto mode:
* `g o` - toggle selected diff hunks
* `g O` - toggle staged
* `g R` - restore change
* `g u` - stage and goto next diff hunk
* `g U` - unstage and goto next diff hunk
This re-introduces the `save_file` and `restore_file_from_disk` agent
tools that were reverted in #44949.
I pushed that original PR without trying it just to get the build off my
machine, but I had missed a step: the tools weren't added to the default
profile settings in `default.json`, so they were never enabled even
though the code was present.
## Changes
- Add `save_file` and `restore_file_from_disk` to the "write" profile in
`default.json`
- Add `Thread::has_tool()` method to check tool availability at runtime
- Make `edit_file_tool`'s dirty buffer error message conditional on
whether `save_file`/`restore_file_from_disk` tools are available (so the
agent gets appropriate guidance based on what tools it actually has)
- Update test to match new conditional error message behavior
Release Notes:
- Added `save_file` and `restore_file_from_disk` agent tools to handle
dirty buffers when editing files
Fixes#44997
## Summary
Optimizes editor rendering when an editor is partially clipped by a
parent container (e.g., a `List`). The editor now only lays out and
renders lines that are actually visible within the viewport, rather than
all lines in the document.
## Problem
When an `AutoHeight` editor with thousands of lines is placed inside a
scrollable `List` (such as in the Agent Panel thread view), the editor
would lay out **all** lines during prepaint, even though only a small
portion was visible. Profiling showed that ~50% of frame time was spent
in `EditorElement::prepaint` → `LineWithInvisibles::from_chunks`,
processing thousands of invisible lines.
## Solution
Calculate the intersection of the editor's bounds with the current
content mask (which represents the visible viewport after all parent
clipping). Use this to determine:
1. `clipped_top_in_lines` - how many lines are clipped above the
viewport
2. `visible_height_in_lines` - how many lines are actually visible
Then adjust `start_row` and `end_row` to only include visible lines. The
parent container handles positioning, so `scroll_position` remains
unchanged for paint calculations.
## Example
For a 3000-line editor where only 50 lines are visible:
- **Before**: Lay out and render 3000 lines
- **After**: Lay out and render ~50 lines
## Testing
Verified the following scenarios work correctly:
- Editor fully visible (no clipping)
- Editor clipped from top
- Editor clipped from bottom
- Editor completely outside viewport (renders nothing)
- Fractional line clipping at boundaries
- Scrollable editors with internal scroll state inside a clipped
container
Release Notes:
- Improved agent panel performance when rendering large diffs.
Release Notes:
- Added project settings schema to the schema_generator CLI. This allows
for exporting the project settings schema as JSON for use in other
tools.
Closes#41994
This PR introduces Element-bounded drag tolerance for Ctrl/Cmd+click in
terminal.
Previously, Ctrl/Cmd+click on terminal links required pixel-perfect
accuracy. Any mouse movement during the click would cancel the
navigation, making it frustrating to click on links, especially on
high-DPI displays or with sensitive mice.
Users can now click anywhere within a clickable element (file path, URL,
hyperlink), drag the cursor anywhere within that same element's
boundaries and release to trigger navigation
Implementation:
- Stores detected element metadata (`text` and `grid_range`) on
Ctrl/Cmd+mouse-down
- Tracks cursor position during drag, preserving click state while
within element bounds
- Verifies element match on mouse-up before triggering navigation
- Uses existing `find_from_grid_point()` for element detection
Before:
[before.webm](https://github.com/user-attachments/assets/ee80de66-998e-4d8e-94d0-f5e65eb06d22)
After:
[after.webm](https://github.com/user-attachments/assets/7c9ddd9e-cfc1-4c79-b62c-78e9d909e6f4)
Release Notes:
- terminal: Fixed an issue where `ctrl|cmd+click` on links was very
sensitive to mouse movement. Clicking links now tolerates mouse movement
within the same clickable element, making link navigation more reliable
---------
Co-authored-by: Ben Kunkle <ben@zed.dev>
Release Notes:
- Fixed double slash in commit URLs
The github_url variable was generating URLs with an extra slash like
"https://github.com//user/repo/commit/xxxx" due to manual string
formatting
of the base_url() result.
Fixed by replacing manual URL construction with the proper
build_commit_permalink() method that uses Url::join() for correct
path handling, consistent with how other Git hosting providers
construct URLs.
Signed-off-by: Xiaobo Liu <cppcoffee@gmail.com>
Closes#40678
The python file below simulates the macros at various timings and can be
run by running:
1. `sudo python3 -m pip install evdev --break-system-packages`
2. `sudo python3 zed_shift_brace_replayer.py`
Checked timings for hold=0.1, =0.01 and =0.001 with the latter two no
longer causing incorrect inputs.
[zed_shift_brace_replayer.py](https://github.com/user-attachments/files/23560570/zed_shift_brace_replayer.py)
Release Notes:
- linux: fixed a race condition where the macros containing modifier +
key would sometimes be processed without the modifier
Related to:
- #44510
- #44407
Previously we were searching for hyperlinks on every scroll, even if Cmd
was not held. With this PR,
- We only search for hyperlinks on scroll if Cmd is held
- We now clear `last_hovered_word` in all cases where Cmd is not held
- Renamed `word_from_position` -> `schedule_find_hyperlink`
- Simplified logic in `schedule_find_hyperlink`
Performance measurements
The test scrolls up and down 20,000x in a loop. However, since this PR
is just removing a code path that was very dependent on the length of
the line in terminal, it's not super meaningful as a comparison. The
test uses a line length of "long line ".repeat(1000), and in main the
performance is directly proportional to the line length, so for
benchmarking it in main it only scrolls up and down 20x. I think all
that is really useful to say is that currently scrolling is slow, and
proportional to the line length, and with this PR it is buttery-smooth
and unaffected by line length. I've included a few data points below
anyway. At least the test can help catch future regressions.
| Branch | Command | Scrolls | Iter/sec | Mean [ms] | SD [ms] |
Iterations | Importance (weight) |
|:---|:---|---:|---:|---:|---:|---:|---:|
| main | tests::perf::scroll_long_line_benchmark | 40 | 16.85 | 712.00 |
2.80 | 12 | average (50) |
| this PR | tests::perf::scroll_long_line_benchmark | 40 | 116.22 |
413.60 | 0.50 | 48 | average (50) |
| this PR | tests::perf::scroll_long_line_benchmark | 40,000 | 9.19 |
1306.40 | 7.00 | 12 | average (50) |
| only overhead | tests::perf::scroll_long_line_benchmark | 0 | 114.29 |
420.90 | 2.00 | 48 | average (50) |
Release Notes:
- terminal: Improved scroll performance
Closes#44825
Release Notes:
- Fixed a case where an incorrect match could be generated in
label_for_completion
---------
Co-authored-by: Kirill Bulatov <mail4score@gmail.com>
This PR introduces a few components for the model selector pickers.
Given we're still maintaining two flavors of it due to one of them being
wired through ACP and the other through the language model registry,
having one source of truth for the UI should help with maintenance
moving forward, considering that despite the internal differences, they
look and behave the same from the standpoint of the UI.
Release Notes:
- N/A
## Summary
This fixes a minor bug I found #44981
- Fix percent-encoded filenames appearing in agent mentions after
message submission.
- Decode file:// paths in MentionUri::parse using the existing
urlencoding crate (already used elsewhere in the codebase).
- Add tests for non-ASCII file URIs.
## Screenshots
<img width="409" height="116" alt="image"
src="https://github.com/user-attachments/assets/32ef033b-6232-47c5-80c7-d5247d5dae88"
/>
## Summary
The diagnostics panel was ignoring the user's `toolbar.breadcrumbs`
setting and always showing breadcrumbs. This makes both
`BufferDiagnosticsEditor` and `ProjectDiagnosticsEditor` check the
`EditorSettings` to determine whether to display breadcrumbs.
## Changes
- `buffer_diagnostics.rs`: Updated `breadcrumb_location` to check
`EditorSettings::get_global(cx).toolbar.breadcrumbs`
- `diagnostics.rs`: Updated `breadcrumb_location` to check
`EditorSettings::get_global(cx).toolbar.breadcrumbs`
This follows the same pattern used by the regular `Editor` in
`items.rs`.
## Test plan
1. Set `toolbar.breadcrumbs` to `false` in settings.json
2. Open a file with diagnostics
3. Run `diagnostics: deploy current file`
4. Verify that breadcrumbs are hidden in the diagnostics panel
Fixes#43020
I had previously added `flex_none` to the Divider and that caused it to
grow beyond the container's width in some cases (project panel, agent
panel's restore to check point button, etc.).
Release Notes:
- N/A
We were previously matching the search on both model name and provider
ID. In most cases, this would yield an okay result, but if you search
for "Opus", for example, you'd see the Sonnet models in the search
result, which was very confusing. This was because we were matching to
both provider ID and model name. "Sonnet" and "Opus" share the same
provider ID, so they both contain "Anthropic" as a prefix. Then, "Opus"
contains the letter P, as well as Anthropic, thus the match.
Now, we're only matching by model name, which I think most of the time
will yield more accurate results.
Release Notes:
- agent: Improved the model search quality in the model picker.
Closes#44306
This PR makes two changes:
- Uses the new `grid_cols_min_content` API. See more here:
https://github.com/zed-industries/zed/pull/44973.
- Changes Markdown table rendering to use a single grid instead of
creating a new grid per row, so column widths stay consistent across
rows.
Release Notes:
- Fixed an issue where Markdown tables wouldn't render in the hover
popover.
Similar to the mode selector in external agents, it will now be possible
to use `shift-tab` to cycle through profiles.
<img width="500" height="384" alt="Screenshot 2025-12-16 at 9 04@2x"
src="https://github.com/user-attachments/assets/11e8824e-9fad-4aab-9e19-53878096db52"
/>
Release Notes:
- Added the ability to use `shift-tab` to cycle through profiles for the
built-in Zed agent.
It was only possible to delete profiles through the `settings.json`, but
now you can do it through the UI:
<img width="500" height="1954" alt="Screenshot 2025-12-16 at 8 42@2x"
src="https://github.com/user-attachments/assets/077ecdf5-1e80-4b70-86c9-177cc3741e77"
/>
Release Notes:
- agent: Added the ability to delete a profile through the "Manage
Profiles" modal.
Required for https://github.com/zed-industries/zed/pull/44712
We started using `grid` for Markdown tables instead of flex. This
resulted in tables having a width of 0 inside popovers, since popovers
are laid out using `AvailableSpace::MinContent`.
One way to fix this is to lay out popovers using `MaxContent` instead.
But that would affect all Markdown rendered in popovers and could change
how popovers look, or regress things.
The other option is to fix it where the problem actually is:
`repeat(count, vec![minmax(length(0.0), fr(1.0))])`. Since the minimum
width here is `0`, laying things out with `MinContent` causes the
Markdown table to shrink completely. What we want instead is for the
minimum width to be the min-content size, but only for Markdown rendered
inside popovers.
This PR does exactly that, without interfering with the `grid_cols` API,
which intentionally follows a TailwindCSS-like convention. See
https://github.com/zed-industries/zed/pull/44368 for context.
Release Notes:
- N/A
### Problem
PR #44411 replaced the `editor::AcceptPartialEditPrediction` action with
`editor::AcceptNextLineEditPrediction` and
`editor::AcceptNextWordEditPrediction`. However, the Linux cursor keymap
wasn't updated to reflect this change, causing a panic on startup for
Linux users.
### Solution
Updated the Linux keymap configuration to reference the new actions
Release Notes:
- N/A
Copying rendered markdown doesn't reliably do anything sensible. If we
copy text from the middle of a bold section, no formatting is copied. If
we copy text at the end, the trailing bold delimiters are copied,
resulting in gibberish markdown. Thus even fixing the associated issue
(so that leading delimeters are reliably copied) won't consistently
produce good results.
Also, as the user messages in the agent panel don't render markdown
anyway, it seems the most likely use case for copying markdown is
inapplicable.
Closes#42958
Release Notes:
- N/A
Hi,
This PR fixes nothing. I just miss the option to open recent projects
quickly upon opening Zed, so I made this. Hope I can see it soon in
Preview channel.
If there is any suggestion, just comment. I will take it seriously.
Thank you!
|ui|before|after|
|-|-|-|
|empty pane|<img width="1571" height="941" alt="Screenshot 2025-12-03 at
12 39 25"
src="https://github.com/user-attachments/assets/753cbbc5-ddca-4143-aed8-0832ca59b8e7"
/>|<img width="1604" height="952" alt="Screenshot 2025-12-03 at 12 34
03"
src="https://github.com/user-attachments/assets/2f591d48-ef86-4886-a220-0f78a0bcad92"
/>|
|new window|<img width="1571" height="941" alt="Screenshot 2025-12-03 at
12 39 21"
src="https://github.com/user-attachments/assets/a3a1b110-a278-4f8b-980e-75f5bc96b609"
/>|<img width="1604" height="952" alt="Screenshot 2025-12-04 at 10 43
17"
src="https://github.com/user-attachments/assets/74a00d91-50da-41a2-8fc2-24511d548063"
/>|
---
Release Notes:
- Added a new value to the `restore_on_startup` setting called
`launchpad`. This value makes Zed open with a variant of the welcome
screen ("the launchpad") upon startup. Additionally, this same page
variant is now also what is displayed if you close all tabs in an
existing window that doesn't contain any folders open. The launchpad
page shows you up to 5 recent projects, making it easy to open something
you were working recently.
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Use the url crate to extract the domain from the verification URI and
construct the appropriate Copilot sign-up URL for GitHub or GitHub
Enterprise.
Release Notes:
- Improved github enterprise (ghe) copilot sign in
Simple documentation PR.
Added information for installing on Windows via winget. Added links from
the main README to relevant sections for both macOS and Windows
Release Notes:
- N/A
First up: I'm sorry if this is a low quality PR, or if this feature
isn't wanted. I implemented this because I'd like to have this
behaviour. If you don't think that this is useful, feel free to close
the PR without comment. :)
My idea is this: I love to pull random models with Ollama to try them.
At the same time, not all of them are useful for coding, or some won't
work out of the box with the context_length set. So, I'd like to change
Zed's behaviour to not show me all models Ollama has, but to limit it to
the ones that I configure manually.
What I did is add an `auto_discover` field to the settings. The idea is
that you can write a config like this:
```json
"language_models": {
"ollama": {
"api_url": "http://localhost:11434",
"auto_discover": false,
"available_models": [
{
"name": "qwen3:4b",
"display_name": "Qwen3 4B 32K",
"max_tokens": 32768,
"supports_tools": true,
"supports_thinking": true,
"supports_images": true
}
]
}
}
```
The `auto_discover: false` means that Zed won't pick up or show the
language models that Ollama knows about, and will only show me the one I
manually configured in `available_models`. That way, I can pull random
models with Ollama, but in Zed I can only see the ones that I know work
(because I've configured them).
The default for `auto_discover` (when it is not explicitly set) is
`true`, meaning that the existing behaviour is preserved, and this is
not a breaking change for configurations.
Release Notes:
- ollama: Added `auto_discover` setting to optionally limit visible
models to only those manually configured in `available_models`
Since the rule is no longer a `style` lint as of
[mid-August](https://github.com/rust-lang/rust-clippy/pull/15454), the
comment mentioning it not being one is outdated and should be removed.
> [!NOTE]
> I kept the severity at `error` for now to avoid rustling feathers.
> If `warn` is preferred, feel free to change it yourself or ask me to
do it - it's only 1 line of code, after all.
Release Notes:
- N/A
Katie Geer