Following up on #49183 which fixed the category filter for remote
extensions.
This PR applies the same category filter logic to **dev extensions**.
Previously, dev extensions were always shown regardless of the selected
category filter (e.g., "Themes", "Languages").
Changes:
- Add `filtered_dev_extension_indices` to track which dev extensions
match the active `provides_filter`
- Add `dev_extension_matches_provides()` helper to map
`ExtensionManifest` fields to `ExtensionProvides` variants
- Update `render_extensions()` and list count to use filtered dev
extension indices
Release Notes:
- Fixed extension category filter not applying to dev extensions in the
extensions panel.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
Closes #ISSUE
Before you mark this PR as ready for review, make sure that you have:
- [ ] Added a solid test coverage and/or screenshots from doing manual
testing
- [ ] Done a self-review taking into account security and performance
aspects
- [ ] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- N/A *or* Added/Fixed/Improved ...
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
Updating content with more detailed information, cross-linking pages
when useful (primarily the new docs for tool permissions), rearranging
information architecture a bit (e.g., putting examples last and
guidelines first, putting text threads vs. threads disclaimer above
everything, etc.), removing outdated images, etc.
Release Notes:
- N/A
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
Follow-up to #49187.
Instead of silently skipping updates when `try_borrow_mut` fails, this
PR defers the update by spawning it on the foreground executor. This
ensures the state change is eventually processed after the current
borrow completes.
Release Notes:
- N/A
The `on_thermal_state_change` and `on_keyboard_layout_change` callbacks
in `App::new_app()` called `borrow_mut()` unconditionally. These
callbacks are invoked asynchronously by macOS via dispatch queues when
the system's thermal or keyboard state changes, which can happen while
the app's RefCell is already borrowed during an update cycle, causing a
panic.
This change uses `try_borrow_mut()` instead. If the borrow fails
(because the app is already borrowed), the callback silently skips the
update - the state can be queried on the next frame.
Fixes [ZED-4WM](https://zed-dev.sentry.io/issues/ZED-4WM)
Closes#49181
Release Notes:
- Fixed a crash on macOS caused by thermal or keyboard layout state
changes occurring during UI updates.
This PR updates the extensions UI to respect the category filter for
already-installed extensions when searching for extensions.
Closes#48628.
Release Notes:
- The extension search will now properly respect the category filter for
extensions that are already installed
([#48628](https://github.com/zed-industries/zed/issues/48628)).
Before you mark this PR as ready for review, make sure that you have:
- [ ] Added a solid test coverage and/or screenshots from doing manual
testing
- [ ] Done a self-review taking into account security and performance
aspects
- [ ] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- N/A
Reverts the whole "project search on type" set of PRs until we figure
out better ways to deal with flickering between search results
appearing.
Release Notes:
- N/A
## Bug
**Sentry:**
[ZED-3W3](https://sentry.io/organizations/zed-dev/issues/7085300207/)
(13 events, first seen 2025-12-03)
`summaries_for_anchors` panics with "cannot seek backward" when
iterating selection anchors whose excerpt locators are no longer in
ascending order after an `update_path_excerpts` call.
### Root cause
`update_path_excerpts` merges new excerpt ranges with existing ones.
When an existing excerpt has no overlap with any new range (e.g.
`existing_range.end < new.context.start`), it is removed from the
excerpts tree but is **not** recorded in `replaced_excerpts`. Its stale
locator persists in the `excerpt_ids` tree.
Later, when a new excerpt is inserted for a different path whose locator
falls between the stale locator and another surviving excerpt,
`summaries_for_anchors` can encounter anchors in an order where it needs
to seek the cursor *backward* — violating the forward-only cursor
invariant.
Concretely, the scenario requires:
1. A path with ≥3 excerpts (E_B1, E_B2, E_B3) with anchors in E_B2 and
E_B3.
2. An `update_path_excerpts` call that keeps E_B1, **removes E_B2** (no
overlap → no `replaced_excerpts` entry), and replaces E_B3 with a new
excerpt N.
3. A new path D inserted between paths B and C, whose excerpt E_D gets a
locator between N and E_B2's stale locator.
4. Resolving the old anchors: E_B2's stale locator seeks past E_D, then
E_B3→N tries to seek backward to N — **panic**.
## Fix
In the two branches of `update_path_excerpts` where existing excerpts
are removed without being absorbed into a new range, map the removed
excerpt to the nearest surviving neighbor (the last kept or inserted
excerpt) in `replaced_excerpts`. This ensures stale anchors always remap
to a valid, correctly-ordered locator.
The two branches are:
- `existing_range.end < new.context.start` — existing excerpt falls
entirely before the next new range
- `(None, Some(...))` — no more new ranges for remaining existing
excerpts
## Verification
- A reproduction test
(`test_cannot_seek_backward_after_excerpt_replacement`) that constructs
the exact crash scenario now passes.
- Full `multi_buffer` test suite (48 tests) passes with no regressions.
- Clippy is clean.
Release Notes:
- Fixed a possible crash when updating excerpts in multibuffers
Closes#42842
Previously, OpenFolderInWsl action assumed a valid path is on the host
and failed to open paths inside the WSL filesystem (accessed through
UNC)
Now the action checks for a WSL path and opens it with the appropriate
distro (skipping the choice modal)
This was manually confirmed to not regress as well as work with WSL
paths.
Code was written by hand, with extra checking by OpenAI Codex.
Release Notes:
- Fixed paths inside the WSL filesystem failing to open when using "open
folder in wsl"
Release Notes:
- Fixed prevent shell command injection in conda environment activation
The conda environment name was directly interpolated into the shell
command without proper escaping, which could allow command injection
if the environment name contained malicious shell metacharacters.
Signed-off-by: Xiaobo Liu <cppcoffee@gmail.com>
on windows 10, When a laptop is plugged into or unplugged from a
monitor, Zed and my using GPUI app will restore from minimized state to
the foreground.
Closes #ISSUE
- [ ] Tests or screenshots needed?
- [ ] Code Reviewed
- [ ] Manual QA
Release Notes:
- N/A *or* Added/Fixed/Improved ...
---------
Co-authored-by: John Tur <john-tur@outlook.com>
This PR uses the recently introduced `CircularProgress` component to
display context window use information. It creates more space for the
message editor controls as well as simplifies the UI a little bit.
Through a tooltip, we communicate the same things we communicated before
(and more, actually, because rules use will be displayed there, too).
Note that this doesn't touch the display of split token use (for models
like GPT and whatnot).
<img width="500" height="484" alt="Screenshot 2026-02-13 at 6 16@2x"
src="https://github.com/user-attachments/assets/831cdfbf-fddd-4a11-8ac6-e9a25609aae9"
/>
Release Notes:
- Agent: Simplified context window use display by using a circular
progress component.
Here's some backstory:
* on macOS, @cole-miller and I noticed that since roughly Oct 2025, due
to some changes to latest macOS Tahoe, for any spawned child process we
needed to reset Mach exception ports
(https://github.com/zed-industries/zed/issues/36754 +
https://github.com/RemiKalbe/zed/commit/6e8f2d2ebe5a93753625a3026aeb996de8cb436b)
* the changes in that PR achieve that via `pre_exec` hook on
`std::process::Command` which then abandons `posix_spawn` syscall for
`fork` + `execve` dance on macOS (we tracked it down in Rust's std
implementation)
* as it turns out, `fork` + `execve` is pretty expensive on macOS
(apparently way more so than on other OSes like Linux) and `fork` takes
a process-wide lock on the allocator which is bad
* however, since we wanna reset exception ports on the child, the only
official way supported by Rust's std is to use `pre_exec` hook
* posix_spawn on macOS exposes this tho via a macOS specific extension
to that syscall `posix_spawnattr_setexceptionports_np` but there is no
way to use that via any standard interfaces in `std::process::Command`
* thus, it seemed like a good idea to instead create our own custom
Command wrapper that on non-macOS hosts is a zero-cost wrapper of
`smol::process::Command`, while on macOS we reimplement the minimum to
achieve `smol::process::Command` with `posix_spawn` under-the-hood
Notably, this changeset improves git-blame in very large repos
significantly.
Release Notes:
- Fixed performance spawning child processes on macOS by always forcing
`posix_spawn` no matter what.
---------
Co-authored-by: Cole Miller <cole@zed.dev>
This PR #47407 broke thinking models for non-zed providers (Nightly
only) because we set `thinking_allowed` to `thinking_enabled`
(previously `true`)
[here](https://github.com/zed-industries/zed/blob/b671c8cdc5a73d49d158a84f386f8fecbfbae65a/crates/agent/src/thread.rs#L2400).
The value of `thinking_enabled` eventually comes from the
`LanguageModel::supports_thinking` function, which returns `false` for
all models except the ones coming from Zed.
This PR implements `thinking_enabled` for other providers.
Before you mark this PR as ready for review, make sure that you have:
- [x] Added a solid test coverage and/or screenshots from doing manual
testing
- [x] Done a self-review taking into account security and performance
aspects
- [x] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- N/A
Closes#47854
Move `disable_ai` from root settings to `ProjectSettingsContent` to
enable per-project AI configuration via `.zed/settings.json`.
- Update settings UI to allow viewing/editing at both user and project
level
- Update editor to check project-level settings for edit predictions and
context menus
- Prevent MCP servers from starting when AI is disabled at project level
Note: SaturatingBool ensures that if user globally disables AI, projects
cannot re-enable it. Projects can only further restrict AI, not grant
it.
Release Notes:
- added support for disabling AI in project settings
---------
Co-authored-by: Ben Kunkle <ben.kunkle@gmail.com>
LSP servers may return symbol names containing newlines. Since outline
panel, breadcrumbs, and project search modal all expect single-line
items, collapse newlines before display.
- Document symbols: collapse newlines to spaces in name
- Workspace symbols: collapse newlines with `↵ ` separator for name and
container_name, preserving visual hint of original structure
Closes #ISSUE
Before you mark this PR as ready for review, make sure that you have:
- [x] Added a solid test coverage and/or screenshots from doing manual
testing
- [x] Done a self-review taking into account security and performance
aspects
- [x] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- N/A
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Closes#46795
This PR aims to correctly detect the hyperlinks in cases like the
following, where box-drawing characters immediately precede the path
(without spaces in between).
```
╭─[Cargo.toml:55:1]
...
╰────
```
The only false negative with the fix in this PR would be that a file
path really contains leading box drawing characters which I think is
very unlikely.
---
Release Notes:
- Fixed an issue where the hyperlinks would not be properly detected in
the terminal if they were preceded by box-drawing characters (like ─, │,
┌, ┐)
Signed-off-by: Charlie-XIAO <yx2436@nyu.edu>
Closes#45135
Since neither the project search nor file finder search was showing
project's root name, including it in the history was unnecessary,
especially when the user had `project_panel.hide_root` set to `true`.
Release Notes:
- Made file finder to respect `project_panel.hide_root` settings
---------
Co-authored-by: Kirill Bulatov <kirill@zed.dev>
We had a call to `BlockMap::unfold_intersecting` that ended up
recomputing the entire block map for the RHS _without_ spacers, only to
throw it away in favor of the version with spacers a few lines down. Now
we only sync each block map once in `set_companion`.
Release Notes:
- Improved performance when toggling from the unified diff to the split
diff.
Co-authored-by: Jakub <jakub@zed.dev>
Co-authored-by: Cameron McLoughlin <cameron.studdstreet@gmail.com>
Fixes#28339
## What
- Normalize cwd-relative path-like targets that include leading `..`
before worktree lookup.
- Update issue #28339 tests to expect `WorktreeExact` for local and
remote resolution.
## Why
- Prevent terminal links like `../foo/bar.txt` from creating invalid
entries; resolve to the correct worktree file instead.
## How to test
- `cargo test -p terminal_view issue_28339 -- --nocapture`
- Manual: open a workspace, run `echo ../foo/bar.txt` from a subdir in
the terminal, then cmd/ctrl-click and confirm the correct file opens.
## Risk
- Low: only changes terminal path-like resolution for `..` paths;
expected to match normalized behavior.
## Checklist
- [x] Tests run (`cargo test -p terminal_view issue_28339 --
--nocapture`)
- [x] Docs updated (not needed)
- [x] Backwards compatibility considered
Release Notes:
- Fixed an issue where relative paths starting with `..` would not resolve correctly when clicking the link in the terminal
---------
Co-authored-by: Ben Kunkle <ben.kunkle@gmail.com>
When the multi-workspace thread restoration feature serializes the
active thread's session ID, it includes empty threads that were never
persisted to the database (because `NativeAgent::save_thread` skips
empty threads). On restart, `AgentPanel::load()` tries to restore this
thread by calling `load_agent_thread()`, which queries the SQLite DB.
Since the empty thread was never saved, the query returns `None`, which
becomes the error: "no thread found with ID: SessionId(...)". This shows
as "Error Loading Zed Agent" / "Failed to Launch" in the UI.
**Fix:**
Before attempting to restore a serialized thread on panel load, query
the DB to verify the thread actually exists. If it doesn't, log an error
and skip restoration (the panel falls back to creating a new thread as
normal).
Closes AI-19
Release Notes:
- Fixed an "Error Loading Zed Agent" error that could appear when
opening a workspace where the previous agent thread was empty.
- [x] Tests or screenshots needed?
- [ ] Code Reviewed
- [x] Manual QA
Release Notes:
- Fixed misalignment of lines in the split diff when using inlay hints.
Accidentally committed the feature flag change in
https://github.com/zed-industries/zed/pull/49100 as I had it turned to
true for local testing. This PR makes it turned off to staff again,
sorry!
Release Notes:
- N/A
Closes#49083
This issue was caused by remote projects on the welcome page sharing the
same ID, as they all used the "Remote Project" label. Additionally, the
tab size was hardcoded to 10, which caused the tab navigation to break.
This PR assigns unique IDs and removes the hardcoded limitation.
- [x] locally checked the click behavior
- [x] self-review done
Release Notes:
- Fixed Welcome page remote projects not responding correctly when
clicked.
Closes #ISSUE
Before you mark this PR as ready for review, make sure that you have:
- [x] Added a solid test coverage and/or screenshots from doing manual
testing
- [x] Done a self-review taking into account security and performance
aspects
- [x] Aligned any UI changes with the [UI
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
Release Notes:
- Fixed an issue where no error was shown when an LLM request exceeded
the maximum tokens supported by the model
This PR adds the ability to, within the Recent Projects modal/popover,
quickly add a project in the recent projects section to the current
workspace. Note that this is currently limited to local projects only.
https://github.com/user-attachments/assets/6b72af11-9c94-45d3-a7df-76869b942727
Release Notes:
- Workspace: Enabled quickly adding a recent project to the current
workspace.
Mistral Vibe, the official coding agent from Mistral, sets the
x-affinity header to the session ID to enable prompt caching. This patch
implements the same mechanism, resulting in a faster agent loop.
Release Notes:
- Added prompt caching for Mistral AI.
Closes#47007
Release Notes:
- Fixed await config worktree before opening settings.json
ensures that the worktree creation is awaited before attempting to open
the settings file.
Signed-off-by: Xiaobo Liu <cppcoffee@gmail.com>
This small change *may* help with #39211, #35025 or #39031 although I'm
not positive it will fix it entirely. At a minimum, it should prevent
sending invalid requests to the Mistral API, which return a 400 error
and can prevent the thread from progressing if it gets stuck in a retry
loop.
This should prevent sending a message like `{ "role": "assistant",
"content": "" }` if for some reason the thread contains a message with
an empty text content and no tool calls.
Some users (including myself) have encountered this with the Devstral 2
models, although it doesn't seem trivial to reproduce:
https://github.com/zed-industries/zed/issues/37021#issuecomment-3728902562
When it occurs, error logs look something like this:
```
2026-01-24T19:53:35-08:00 ERROR [crates/agent/src/thread.rs:984] invalid type: string "Failed to connect to Mistral API: 400 Bad Request {\"object\":\"error\",\"message\":\"Assistant message must have either content or tool_calls, but not none.\",\"type\":\"invalid_request_assistant_message\",\"param\":null,\"code\":\"3240\"}", expected struct EditFileToolOutput
2026-01-24T19:53:35-08:00 ERROR [crates/agent/src/thread.rs:984] invalid type: string "Failed to connect to Mistral API: 400 Bad Request {\"object\":\"error\",\"message\":\"Assistant message must have either content or tool_calls, but not none.\",\"type\":\"invalid_request_assistant_message\",\"param\":null,\"code\":\"3240\"}", expected struct EditFileToolOutput
2026-01-24T19:53:35-08:00 ERROR [crates/agent/src/thread.rs:984] invalid type: string "Failed to connect to Mistral API: 400 Bad Request {\"object\":\"error\",\"message\":\"Assistant message must have either content or tool_calls, but not none.\",\"type\":\"invalid_request_assistant_message\",\"param\":null,\"code\":\"3240\"}", expected struct EditFileToolOutput
```
Release Notes:
- Fixed agent sometimes sending invalid messages to Mistral API
Closes #N/A (no existing issue - implemented to enable image input for
Bedrock models)
This PR enables the "@" image mention feature for Bedrock models that
support vision capabilities.
**Changes:**
- Added `supports_images()` method to Bedrock `Model` enum
- Wired up image support in the Bedrock language model provider
- Added `MessageContent::Image` handling to convert base64 images to
Bedrock's expected format
- Added tool result image support
**Supported models:** Claude 3/3.5/4 family, Amazon Nova Pro/Lite, Meta
Llama 3.2 Vision, Mistral Pixtral
Release Notes:
- Added image input support for Amazon Bedrock models with vision
capabilities
Closes#48732
- [X] Tests or screenshots needed?
- [X] Code Reviewed
- [X] Manual QA
The fix enables interactive mode during task execution, resolving the
Conda activation error. However, enabling this mode adds some overhead.
In my environment, the task startup time increased by approximately 1
seconds due to the loading of ~/.zshrc.
[录屏 2026-02-08
23-49-53.webm](https://github.com/user-attachments/assets/4064e366-b0d2-4a8c-9599-c4108e76af9f)
Release Notes:
- Fixed conda activation error appears during task execution.
Follow-up to https://github.com/zed-industries/zed/pull/41111.
PR #41111 introduced combined injection handling, but cursor language
queries still relied on layer range selection alone. For combined
injections this can surface a language from the outer combined layer
even when the cursor is outside that language's actual included
sub-range. So, we just need to add sub-range filtering based on
anchor-aware boundary checks in the same way we did in the previous PR.
That means: apply it in `Buffer::language_at`,
`Buffer::languages_at`, and `BufferSnapshot::language_scope_at`. All
places that rely on the described behavior.
I also added some additional test cases for the `HTML+ERB` lang to
verify language resolution for HTML and Ruby positions.
Thank you!
Closes https://github.com/zed-industries/zed/issues/48358
Release Notes:
- Respect combined injection sub-ranges for language queries
Smit Barmase