Reverts the editor's paragraph navigation behavior that was changed in
#47734. Whitespace-only lines are now treated as paragraph boundaries
again for non-vim mode users.
Vim mode retains its own implementation where only truly empty lines
are paragraph boundaries.
Release Notes:
- Fixed editor paragraph navigation to treat whitespace-only lines as
paragraph boundaries again
---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: dino <dinojoaocosta@gmail.com>
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
- [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
Release Notes:
- Improved Bedrock error messages: region-locked models ask the user to
try a different region, rate limits and access errors are reported
cleanly instead of as raw API responses
- Streamlined Bedrock model list to 39 curated models
- Fixed API errors when using non-tool models in agent threads
---------
Co-authored-by: Ona <no-reply@ona.com>
It now creates multi-turn conversation, where first two messages are the
original teacher prompt and output.
This way, we'll have changes in teacher prompt automatically applied to
repair without having to explicitly sync them.
Release Notes:
- N/A
Other changes:
- Changed tokenization to more code-aware tokenization from split-commit
- Fixed word-diff implementation which was inefficient and sometimes
incorrect
Release Notes:
- N/A
Our strategy for keeping excerpts in sync between the two sides of the
split diff is generally
- mutate the excerpts on the RHS
- pull out the new excerpts for the affected path from the RHS
- translate those ranges to the LHS
- update the LHS excerpts with those ranges
But this can break down when the translated ranges overlap on the LHS
despite the original ranges not overlapping on the RHS. This can happen,
for example, when a large deletion in a file crosses several excerpt
boundaries. It's rare because normally excerpt boundaries in the project
diff and branch diff are adjusted to keep entire diff hunks in view, but
we defer this for dirty files, so it's not hard to trigger it when using
the agent or editing an individual file while the diff is open in
another tab.
Release Notes:
- Fixed a panic that could occur when editing files while the project
diff or branch diff was open.
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
Closes#49172
- Fixes a regression on X11 where content size was divided by the scale
factor twice after the wgpu migration, causing half-sized rendering on
HiDPI screens.
- Aligns X11 [content_size()] behavior with Wayland by returning logical
pixel bounds directly.
Video :
[Screencast from 2026-02-14
18-58-36.webm](https://github.com/user-attachments/assets/cf4edf3b-7b93-4ded-bf46-e900be44b22b)
Release Notes:
- N/A
This PR removes some code left over from how we used to do impersonation
for local development.
The local development impersonation path is separate now, so we don't
need all of this plumbing in Collab.
Release Notes:
- N/A
This PR removes the `GET /rpc_server_snapshot` endpoint from Collab, as
it is not being used (and also didn't seem to be functional).
Release Notes:
- N/A
Re-lands https://github.com/zed-industries/zed/pull/48891, which was
reverted due to conflicts with multi-workspace.
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
Release Notes:
- N/A
This PR adds a button in the Git Panel to open the `git: branch diff`
view. The button only shows if the staging area is clean/no changes and
if the branch is not main/master. I've been using this action so much
these days and we should expose it on the UI so it's more discoverable.
<img width="500" height="1736" alt="Screenshot 2026-02-16 at 6 30@2x"
src="https://github.com/user-attachments/assets/0b037811-2167-419f-b63f-9bdb12e4352e"
/>
Release Notes:
- N/A
Second attempt at fixing
https://github.com/zed-industries/zed/pull/49047, this also removes some
of the sorting hacks as they should ideally not be necessary anymore
after this fix
Release Notes:
- N/A *or* Added/Fixed/Improved ...
This change improves performance of project diff in that:
* scrolling in split view for very large diffs (think chromium repo with
`git reset HEAD~1000`) is now very smooth on macOS and fairly smooth on
Linux
* switching from split to unified is very smooth on macOS, and fairly
smooth on Linux
There still remains the case of (severe) hangs when switching from
unified to split however, but it will be addressed in a follow-up PR.
Anyhow, here's the screenshot of the Instruments.app capture of opening
chromium repo in Zed in split view, scrolling a little, moving to
unified, scrolling some more, and moving back to split. Prior to this
change, split -> unified would cause a severe hang, whereas now it's a
hang and thus feels much smoother already (without Instruments profiling
is barely visible). Unified -> split severe hangs are still there but
don't last as long.
<img width="2301" height="374" alt="Screenshot 2026-02-16 at 5 46 23 PM"
src="https://github.com/user-attachments/assets/f687f8d4-cffd-47f1-ada1-f6c4d3ac3cd4"
/>
Release Notes:
- Improved project diff performance when opening very large
diffs/repositories.
### Description
Related Discussions: #44499, #35742, #31851
Display cost multiplier for GitHub Copilot models in the model selectors
(Both in Chat Panel and Inline Assistant)
<img width="436" height="800" alt="image"
src="https://github.com/user-attachments/assets/c9ebd8fa-4d55-4be8-b3e1-f46dbf1f0145"
/>
### Some technical notes
Although this PR's primary intent is to show the cost multiplier for
GitHub Copilot models alone, I have included some necessary plumbing to
allow specifying costs for other providers in future. I have introduced
an enum called `LanguageModelCostInfo` for showing cost in different
ways for different models. Now, this enum is used in `LanguageModel`
trait to get the cost info.
For now to begin with, in `LanguageModelCostInfo`, I have specified two
ways of pricing: Request-based (1 Agent request - GitHub Copilot uses
this) and Token-based (1M Input tokens / 1M Output tokens). I had
initially thought about adding a `Free` type, especially for Ollama but
didn't do it after realizing that Ollama has paid plans. Right now, only
the Request-based pricing is implemented and used for Copilot models.
Feel free to suggest changes on how to improve this design better.
Release Notes:
- Show cost multiplier for GitHub Copilot models
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
This PR removes the `cloud-thinking-effort` feature flag to ship the
thinking effort UI for the Zed provider.
Release Notes:
- Added support for controlling thinking effort levels with supported
models using the Zed provider.
It is possible, when the agent is streaming the tool call, that we will
get a permission request on an existing terminal and we need to render
that.
When Zed controls the terminal this won't happen, but when we are just
rendering the output, it can happen and we need to show it.
Release Notes:
- agent: Fix permission options not rendering on terminal commands
Bedrock Converse API requires toolConfig whenever messages have tool use
or result blocks. PR #33174 handled the case where tools are defined but
tool_choice is None. This change addresses another case: the tools array
is empty (e.g. thread summarisation) but messages still have tool blocks
from conversation history. A placeholder tool satisfies the requirement.
Testing:
Tested manually by:
1. Starting a conversation with Bedrock that uses tools
2. Triggering thread summarisation
3. Confirming summarisation now succeeds instead of failing with an API
error
Release Notes:
- Fixed Bedrock thread summarization failing when conversation had tools
Fixes#49062
Updates `crates/agent_ui/src/inline_assistant.rs` to ensure the
TerminalPanel correctly syncs the assistant_enabled
state on initialization. This fixes the issue where "Add to Agent
Thread" was missing in worktree-less windows.
## Video:
[Screencast from 2026-02-13
10-31-43.webm](https://github.com/user-attachments/assets/bee0df67-0ecf-460c-8123-a72c4790c719)
## Release Notes
- Fixed an issue where "Add to Agent Thread" was missing from the
terminal context menu in windows without open folders.
## Summary
Fixes https://github.com/zed-industries/zed/issues/48684
- Await `parsing_idle()` after `set_language()` in both
`Diff::finalized()` and `PendingDiff::finalize()` so that tree-sitter
parsing completes before buffer snapshots are taken for excerpts
- Without this, the 1ms sync parse timeout almost always expires for
non-trivial files, causing excerpts to be created with empty syntax
trees and missing syntax highlighting
## Root Cause
When the agent panel creates or finalizes a diff, it creates a new
`Buffer`, calls `set_language()` (which triggers async tree-sitter
parsing), then immediately proceeds to take buffer snapshots and create
excerpts **without waiting for parsing to complete**. The sync parse
timeout is only 1ms in production (`buffer.rs:1138`), so large files
almost always fail the sync parse and fall back to async parsing — but
no one awaits the async parse before building the diff.
## Fix
Add `buffer.update(cx, |buffer, _| buffer.parsing_idle()).await` after
`set_language()` in both code paths (`Diff::finalized()` and
`PendingDiff::finalize()`). This is the same pattern already used in
`buffer_diff.rs:1728-1731`.
## Test plan
- [x] `cargo test -p acp_thread` — all relevant tests pass
- [x] `./script/clippy` — no warnings
- [x] Manual test: open agent panel, ask it to generate a Python file
with multiline strings (`"""`), verify syntax highlighting is correct
after diff is finalized
Release Notes:
- Fixed missing syntax highlighting for multiline strings in agent panel
diffs
Add StreamEnded variant so the client can distinguish between a stream
that the cloud ran to completion versus one that was interrupted (see
CLO-258). **That logic is to be added in a follow up PR**.
Add an Unknown fallback with #[serde(other)] for forward-compatible
deserialization of future variants.
The client advertises support via a new
x-zed-client-supports-stream-ended-request-completion-status header. The
server will only send the new variant if that header is passed. Both
StreamEnded and Unknown are silently ignored at the event mapping layer
(from_completion_request_status returns Ok(None)).
Part of CLO-264 and CLO-266; cloud-side changes to follow.
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
Changes the AcpThreadHistory behavior to only load the full paginated
list if we are in the history view, and only reloads the first page
otherwise. And it also updates the first page in way fewer cases to
reduce load on the agents.
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/46926
## Description:
- Fixes the missing right border on pinned tabs when
`show_pinned_tabs_in_separate_row` is enabled.
The two-row tab bar layout was missing the border element that visually
separates the pinned tabs row. This border was present in the single-row
layout but was not added when implementing the two-row layout.
## Steps to reproduce
1. Enable `"tab_bar": { "show_pinned_tabs_in_separate_row": true }` in
settings
2. Open multiple files
3. Pin at least one tab (right-click → Pin Tab)
4. Notice pinned tabs are missing right-hand side border
**Before (bug):**
<img width="1060" height="186" alt="image"
src="https://github.com/user-attachments/assets/7d18f24d-a51b-4f13-8410-a15fa92e1bb3"
/>
**After (fixed):**
<img width="863" height="69" alt="image"
src="https://github.com/user-attachments/assets/3354e809-b298-49a1-b16a-871caef67176"
/>
## Test plan
- [x] Follow reproduction steps above and verify the border now appears
- [x] Verify border appears immediately (not just after drag-and-drop)
- [x] Verify single-row layout still works correctly when setting is
disabled
- [x] Added automated test `test_separate_pinned_row_has_right_border`
### Additional fix: Drag-and-drop to pinned tabs bar
During implementation, discovered that we couldn't drag tabs to the end
of the pinned tabs bar. This PR also adds:
- A drop target at the end of the pinned tabs row
- Proper pinning behavior when dropping unpinned tabs to the pinned area
- Keeps the dragged tab active after drop
Release Notes:
- Fixed the missing right border on pinned tabs when
`show_pinned_tabs_in_separate_row` is enabled.
- Fixed drop target at the end of the pinned tabs row
- Fixed pinning behavior when dropping unpinned tabs to the pinned area
- Fixed case when the dragged tab was not active after drop (when enable
`"tab_bar": { "show_pinned_tabs_in_separate_row": true }` in settings)
---------
Co-authored-by: Joseph T. Lyons <JosephTLyons@gmail.com>
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Closes#47678
When using mixed sort mode in the project panel, a folder and file with
the same name but different case (e.g., `hello` folder and `Hello.txt`
file) would sort incorrectly. The file could appear between an expanded
folder and its contents.
The issue was in `compare_rel_paths_mixed`: the tie-breaker logic used
case-sensitive comparison (`a == b`) to decide directory-before-file
ordering, but `natural_sort_no_tiebreak` already considers entries equal
case-insensitively. Changed to use `eq_ignore_ascii_case` to match.
Release Notes:
- Fixed project panel mixed sort mode ordering incorrectly when a file
and folder share the same name with different casing.
GeoJSON is a popular geospatial data interchange format, standardised as
[RFC 7946](https://datatracker.ietf.org/doc/html/rfc7946). Because all
GeoJSON files are valid JSON files, they can be highlighted using Zed's
existing JSON language support.
This change adds `geojson` as a recognised path suffix, so GeoJSON files
automatically open as JSON and get the standard JSON syntax
highlighting.
Before and after screenshots:
<img width="1748" height="755" alt="geojson"
src="https://github.com/user-attachments/assets/40199248-1ce5-451e-9200-5b2f012b865f"
/>
Release notes:
- Added automatic syntax highlighting for GeoJSON files
Derive collapsed state from `Editor.has_any_buffer_folded` instead of
tracking it separately, removing redundant `ResultsCollapsedChanged` event
and stale `is_collapsed`/`results_collapsed` fields.
Closes#48734
Release Notes:
- Fixed collapse/expand all button in buffer search and project search
not syncing correctly when toggling individual file sections
---------
Signed-off-by: Benjamin <5719034+bnjjj@users.noreply.github.com>
Co-authored-by: dino <dinojoaocosta@gmail.com>
It's possible that a given project/workspace has symlinks that go
outside the current directories. In order to alert the user of some
potential files the agent is accessing they may not be expecting, we ask
for permission first.
Release Notes:
- agent: Prompt for permission if the agent tries to access files whose
symlinks resolve outside the current workspace
When a tool's JSON response fails to parse, the system would:
1. Create a `LanguageModelToolResult` with the error
2. Add it to `pending_message.tool_results`
3. **Never add the corresponding `ToolUse` to
`pending_message.content`**
This left an orphaned `tool_result` that would be sent to the LLM API
without a matching `tool_use` block, causing the provider to reject the
entire request with an error like:
```
messages: Assistant message must contain at least one content block, if
immediately followed by a user message with tool_result
```
The issue was in `handle_tool_use_json_parse_error_event()`. It created
and returned a `LanguageModelToolResult` (which gets added to
`tool_results`), but **failed to add the corresponding `ToolUse` to the
message `content`**.
This asymmetry meant:
- `pending_message.content`: [] (empty - no ToolUse!)
- `pending_message.tool_results`: {id: result}
When `AgentMessage::to_request()` converted this to API messages, it
would create:
- Assistant message: no tool_use blocks ❌
- User message: tool_result block ✅
APIs require tool_use and tool_result to be paired, so this would fail.
**Without this fix, the conversation becomes permanently broken** -
every subsequent message in the thread fails with the same API error
because the orphaned tool_result remains in the message history. The
only recovery is to start a completely new conversation, making this a
particularly annoying bug for users.
Modified `handle_tool_use_json_parse_error_event()` to:
1. **Add the `ToolUse` to `pending_message.content`** before returning
the result
2. Parse the raw_input JSON (falling back to `{}` if invalid, as the API
requires an object)
3. Send the `tool_call` event to update the UI
4. Check for duplicates to avoid adding the same `tool_use` twice
This ensures `tool_use` and `tool_result` are always properly paired.
Added comprehensive test coverage for
`handle_tool_use_json_parse_error_event()`:
- ✅ Verifies tool_use is added to message content
- ✅ Confirms tool_use has correct metadata and JSON fallback
- ✅ Tests deduplication logic to prevent duplicates
- ✅ Validates JSON parsing for valid input
## Manual Testing
To reproduce and test the fix:
1. Install the test MCP server:
```bash
cargo install --git https://github.com/dastrobu/mcp-fail-server
```
3. Add to Zed settings to enable the server:
```json
{
"context_servers": {
"mcp-fail-server": {
"command": "mcp-fail-server",
"args":[]
}
}
}
```
4. Open the assistant panel and ask it to use the `fail` tool
5. Without the fix: The conversation breaks permanently - every
subsequent message fails with the same API error, forcing you to start a
new thread
<img width="399" height="531" alt="image"
src="https://github.com/user-attachments/assets/533bdf40-80d3-4726-a9d9-dbabbe7379e5"
/>
7. With the fix: The error is handled gracefully, displayed in the UI,
and the conversation remains usable
<img width="391" height="512" alt="image"
src="https://github.com/user-attachments/assets/73aa6767-eeac-4d5d-bf6f-1beccca1d5cb"
/>
The mcp-fail-server always returns an error, triggering the JSON parse
error path that previously caused orphaned tool_result blocks.
Release Notes:
- Fixed an issue where errors could occur in the agent panel if an LLM
emitted a tool call with an invalid JSON payload
---------
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Delete all the hacks and do things properly instead.
Closes https://github.com/zed-industries/zed/issues/45574
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
Release Notes:
- N/A
Lukas Wirth