Closes#53266
When a tool call awaiting permission is below the viewport, it’s easy to
miss why the agent has stopped working. This PR adds a floating row at
the bottom of the panel that mirrors the pending permission request.
It reuses the same permission prompt, so you can allow or deny it and
see the raw input without scrolling to the end of the thread. It offers
a "Scroll to" button to jump back to the inline prompt in the thread.
Once the inline prompt is in view, the floating row disappears. If
multiple tool calls need permission, including a subagent’s, they’re
shown one at a time so rows don’t stack on top of each other.
<img width="508" height="181" alt="image"
src="https://github.com/user-attachments/assets/7d5fedd6-bc4d-4674-8ea2-5c8f2ed3aff6"
/>
Release Notes:
- Improved visibility of pending tool call confirmations in the Agent
Panel when its scrolled out of view.
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
- don't fail to parse json when Claude ignores the instructions and
returns fenced json
- don't fail to search for issues by area labels when Claude ignores the
instructions and returns prose instead of comma-separated list
- don't mark workflow runs as successful when json parsing blew up or
posting the comment failed
Release Notes:
- N/A
When restarting Zed, I hit a bug where all Git operations were hung. I
realized that there was a hanging git process running `git cat-file
--batch-check=%(objectname)`. The process was waiting on stdin. This was
surprising, because [the
code](e2bbdb19b6/crates/git/src/repository.rs (L1665-L1709))
that spawns this process explicitly closes the pipe that is attached to
the process's stdin after writing a list of ref names.
Using Claude, I found that this could be caused by that pipe file
descriptor being cloned due to file descriptor inheritance when another
child process is `exec`'d while that stdin pipe is open. The fix is to
enhance our Darwin process spawning layer to set the close-on-exec flag
for the pipe file descriptors, so that they are not inherited by child
processes spawned using code paths that don't set
`POSIX_SPAWN_CLOEXEC_DEFAULT`.
Release Notes:
- Fixed a bug on macOS where Git operations could be blocked depending
on the timing of spawning child processes.
This PR changes the git store's `compute_snapshot`, which runs to update
state that depends on the contents of `.git`, to degrade gracefully when
fetching individual pieces of state fails. For example, when fetching
the list of branches fails, instead of returning early from the function
(leaving the previous git state snapshot in place with stale state), we
continue with an empty list of branches. This prevents failures of
individual git commands from making the entire git UI get stuck
indefinitely.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- Fixed an issue where failing to fetch branches using the git CLI would
prevent other git-related state from being updated.
When viewing the diff for a commit where one of the files has a large
number (hundreds or thousands) of diff hunks, we can get a hang on the
main thread, since there's a large amount of work to recompute diff
transforms for thousands of excerpts. This PR makes it so that we batch
the addition of excerpts, with yield points in between batches so the UI
remains responsive.
We should have something similar for the project diff, but that's a bit
trickier; will address it in a separate PR.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Improved performance when opening the commit view with files that have
many diff hunks.
Following up on our recent extension outage, this PR adds an action to
quickly visit our status page (which with the help of @\gaauwe was
recently updated to a more up to date design 🎉 ).
While already invokable through the command palette, we can later use
this to also add buttons where appropriate for a quick lookup of our
current status.
Release Notes:
- Added the `zed::OpenStatusPage` action to be able to easily visit the
Zed status page.
cc @cole-miller
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Fixed branch enumeration when a broken Git ref prevents commit
metadata from being read.
---------
Co-authored-by: Cole Miller <cole@zed.dev>
In addition to the user-facing suggestions with potential duplicates,
the bot will now also have a separate collapsible block in the same
comment that's meant more for the triagers. This block will list the
duplicates we're less sure about as well as recently closed issues that
seem related to the issue at hand.
This commit updates the bot to V3 for the tracking-effectiveness
purposes and fixes some leftovers from the previous dupe bot change
(activating the bot for issues with no type).
It also improves the reliability of the script a bit, wrapping some
calls to the GitHub API into retries.
Release Notes:
- N/A
Summary:
- Added a rename action for agent threads in the sidebar.
- Persisted renamed thread titles and kept open thread views in sync.
Release Notes:
- Improved agent threads by allowing them to be renamed directly from
the sidebar.
When shift is held during a single click, extend the existing selection
from its anchor point to the clicked position instead of starting a new
selection.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Added markdown support shift+click to extend text selection
We regularly get a certain amount of 'blank' issues on github — those
created bypassing all our templates — and since a fair share of them are
bug reports, it would be helpful for the person triaging if the
duplicates bot has left suggestions on these issues too.
Release Notes:
- N/A
When generating a training or evaluation example with `ep split-commit`,
the cursor sampling logic becomes:
1. 80% chance of cursor being at the end of the source patch
2. 20% chance of cursor being at the beginning of the target patch
3. 20% chance of adding a jitter offset (same line, ±5 columns for now)
Release Notes:
- N/A
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Resolves the issues described at
https://github.com/pop-os/cosmic-text/releases/tag/0.19.0
This fix addresses the characters '\u{001C}' | '\u{001D}' | '\u{001E}' |
'\u{0085}' | '\u{2029}', which can cause the Editor gpui-component to
crash, as these characters can cause the 'assert "left == right"'.
Release Notes:
- N/A
The "all_editor" option for the `minimap.display_in` key is invalid. The
correct option is "all_editors".
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- N/A
This is useful when running tests for a large number of iterations
noninteractively
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- N/A
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- Fixed an issue where edit prediction previews that appeared in the
diff popover could be occluded by open docks or the sidebar on the right
side of the editor
Self-Review Checklist:
- [X] I've reviewed my own diff for quality, security, and reliability
- [X] Unsafe blocks (if any) have justifying comments
- [X] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [X] Tests cover the new/changed behavior
- [X] Performance impact has been considered and is acceptable
Closes#57463
Release Notes:
- Fixed utf-8 parsing issues when loading skill frontmatter with
multi-codepoint graphemes (such as an emoji) crossing file chunk load
boundaries.
Format `read_file` tool output in `cat -n` style: each line is prefixed
with its line number right-aligned in a 6-character field, followed by a
single tab, followed by the line's original content (newlines preserved,
including CRLF). Numbering reflects the actual file lines, so a ranged
read starting at line 42 emits `42` for its first line, not `1`.
For large files, content returned by `get_buffer_content_or_outline` is
**not** prefixed:
- The symbol outline path already conveys structure via its `[L100-150]`
annotations.
- The truncated first-1KB fallback (used when a file exceeds
`AUTO_OUTLINE_SIZE` and has no parseable outline) is wrapped in a
synthetic `# First 1KB of …` header, so its lines don't correspond to
real file line numbers.
Both cases are reported via `BufferContent::is_synthetic` (renamed from
`is_outline`).
Also updates the `edit_file` tool's input doc to describe the prefix
format and tell the model to strip it before constructing `old_text` /
`new_text`, preserving the original indentation that appears after the
tab.
Updates how we render `read_file` tool call outputs in the UI
(screenshots included in comments below).
Also fixes an existing bug where `read_file` tool call outputs would not
re-render their content code block when an older thread was restored
(the tool's `replay` hook was missing).
Closes AI-226
Release Notes:
- Improved how `read_file` tool output renders in the agent panel, with
a line-number gutter, and fixed it not re-rendering on restored threads
---------
Co-authored-by: zed-zippy[bot] <234243425+zed-zippy[bot]@users.noreply.github.com>
If the agent server supports it, we pass along the multiple directories.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes#54480
Release Notes:
- acp: External agents that support it now have access to all working
directories in a project.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- acp: Add Logout flow for agents that support it.
The circular progress in the agent panel's message editor would
previously display the number of default user rules auto-embedded into
the thread. However, given default rules are all migrated to a global
AGENTS.md file, that feature doesn't make a lot of sense anymore. So,
this PR fixes it by adding a button that opens up the global file when
it exists.
Release Notes:
- N/A
Tiny changes: improving keyboard nav by fixing the tab order, and
improving styles by making the dropdown take just the width of the label
instead of a fixed width.
Release Notes:
- N/A
Quick follow-up to https://github.com/zed-industries/zed/pull/57322. We
weren't removing the pending notification when re-visiting the project
that contains the pending thread (only while the project is still
collapsed, though).
Release Notes:
- N/A
## Summary
- Hide deleted skills immediately in Settings while deletion completes
- Refresh the skill index after creating a skill so Settings updates
without reopening
Closes AI-299
Release Notes:
- Fixed skill management so newly created and deleted skills update in
Settings immediately.
- Add a separate `git_commit_buffer_font_size` setting, defaulting to
`12px` (the previous default before it was changed to use the buffer
font size)
- Add in-memory buffer font size overrides for zooming the commit modal
and in-panel editor
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Added a `git_commit_buffer_font_size` setting and made the in-panel
and modal commit message editors zoomable.
When the agent mentions a file path inside `backticks` (e.g. ``
`src/main.rs` `` or `` `src/main.rs:42` ``), the rendered code span now
becomes a clickable link in the agent panel. Clicking opens the
referenced file in the workspace, jumping to the right line and column
when present.
## How it works
- **Shared path resolution.** Extracted `OpenTarget` and the
workspace/worktree resolution logic out of
`terminal_view::terminal_path_like_target` into a new
`workspace::path_link` module so both the terminal and the agent panel
can use the same code. Includes a `sanitize_path_text` helper ported
from the terminal's URL/punctuation handling. Pure refactor — terminal
behavior is unchanged.
- **`markdown` crate hook.** Added
`MarkdownElement::on_code_span_link(callback)`. When the callback
returns `Some(url)` for a given code span's contents, the existing
`push_link` machinery wires up cmd-hover, hit testing, and the existing
`on_url_click` callback. When it returns `None`, the code span renders
as before. The hook is opt-in, so `markdown` stays workspace-agnostic.
- **Agent panel wiring.** `render_agent_markdown` constructs an
`AgentCodeSpanResolver` that snapshots the project's visible worktree
entries plus their file extensions. `try_resolve` does a cheap
synchronous heuristic check (path must contain `/`/`\` or end in an
extension present in the workspace, can't be a URL, can't be all digits,
etc.) and then looks the candidate up in the per-worktree
`HashSet<Arc<RelPath>>`. On a hit it returns a `MentionUri::File` or
`MentionUri::Selection` URI, which the existing `thread_view::open_link`
already knows how to open at the right line.
## Edge cases handled
- Code spans inside fenced code blocks stay plain (gated on
`builder.code_block_stack.is_empty()`, matching how regular markdown
links behave).
- Trailing prose punctuation (`` `src/main.rs.` ``) is stripped before
lookup.
- Identifiers like `` `String` ``, `` `await` ``, `` `npm run dev` ``
stay plain — they don't pass the path-like heuristic.
- Cross-platform path separators handled via the per-worktree
`PathStyle`.
## Tests
- `crates/markdown` — unit test asserting code spans become links when
the callback returns `Some`, and stay plain when it doesn't.
- `crates/agent_ui` — unit test for `AgentCodeSpanResolver::try_resolve`
covering hits with and without a `:line` suffix, misses, identifiers,
and trailing punctuation.
- Existing `terminal_view` tests cover the moved resolution code
(unchanged behavior).
## Notes
- There's currently a temporary `log::info!` in
`AgentCodeSpanResolver::try_resolve` that reports per-call worktree-walk
timing and a cumulative total. Kept in for now to verify the feature
isn't being called excessively during streaming renders. Can be removed
before merge.
- Resolution is sync-only against worktree entries; absolute paths
outside the workspace are not resolved (would require an async re-render
path).
Closes AI-277
Release Notes:
- Made file paths in `backticks` clickable in the agent panel; clicking
opens the referenced file at the given line when present.
Closes AI-298
This PR adds the first step towards allowing to reorganize the threads
sidebar. Drag and drop should be supported in the near future, maybe
even replacing this entirely:
<img width="700" alt="Screenshot 2026-05-21 at 6 44@2x"
src="https://github.com/user-attachments/assets/db420466-2323-474b-ba41-17eb4da2cf84"
/>
Release Notes:
- Sidebar: Added the ability to reorder projects by moving them up and
down through the ellipsis menu.
Whenever there are subpages in the settings UI, a breadcrumb is
displayed. However, we weren't displaying the scope as part of the
breadcrumb, which can be relevant if you're wondering why a certain
information you expected to see here isn't being displayed. It might
just be that it is displayed in the _project_ scope instead of in the
_user_ scope. Therefore, this PR adds the scope in the breadcrumb:
| Before | After |
|--------|--------|
| <img width="1348" height="290" alt="Screenshot 2026-05-21 at 4 38
2@2x"
src="https://github.com/user-attachments/assets/1b64ba98-8e7e-41a7-9b13-19722bdbe093"
/> | <img width="1348" height="290" alt="Screenshot 2026-05-21 at 4
38@2x"
src="https://github.com/user-attachments/assets/b5deb091-0617-42ac-bb4a-b8ba00ec386c"
/> |
Release Notes:
- Started to display the setting scope (user or project) in the Settings
Editor for better contextualization of settings subpages.
Closes AI-280
When navigating between projects opened in the same window through the
recent projects picker, the confirm action would always default to
picking the main worktree in a project. So if you were in a Git worktree
in project A, switched to project B, and then back to project A, when
coming back, you wouldn't be in the Git worktree you were in before. The
fix is done through matching by project group key instead of file system
paths. Since both the main and linked worktrees share the same key, it
will just find the previously active workspace and activate it.
Release Notes:
- Fixed a bug where navigating through open projects in the same window
through the recent projects would always default to the main worktree of
a given project, instead of activating the last active
worktree/workspace.
Closes AI-291
This PR adds some new menu items in the agent panel's ellipsis menu to
make it easier to 1) create a new skill, 2) manage all existing skills
through the settings window, 3) open project and global AGENTS.md file,
if it exists.
Minor UI note: I'm adding an ellipsis character to the menu items in
which you need to do something else in a different view to fulfill the
item's label. For example, "Create Skill…" takes the ellipsis because
clicking on the menu item itself doesn't create the skill, but it takes
you to the place where you'll do that.
<img width="600" alt="Screenshot 2026-05-21 at 10 32@2x"
src="https://github.com/user-attachments/assets/1bfdd423-bd75-4c62-97ba-84130035ede8"
/>
Release Notes:
- N/A
Closes AI-285
Similar to how we display whether there are running threads or a thread
waiting for permission in the collapsed version of the project's header
in the sidebar, I was missing the "unread" state from being shown. I had
to change the approach here as to how we extract this information
because the previous method was relying on observing the state of the
list entries within the sidebar at every `rebuild_contents` run, and
given there aren't any list entires when the project is collapsed, it
wouldn't work.
Release Notes:
- Agent: Added the notification indicator on collapsed project headers
in the sidebar when a thread completes.
Closes AI-295
This PR adds a skills submenu within the "add context" menu in the agent
panel's message editor. This will hopefully be yet another way to find
skills in the app.
<img width="600" alt="Screenshot 2026-05-21 at 11 24@2x"
src="https://github.com/user-attachments/assets/43652081-3929-4ca0-a32b-464077d84dd5"
/>
Release Notes:
- N/A
Clicking a built-in skill mention while connected to a remote project
(SSH or collab) caused Zed to abort with `called create_local_buffer on
a remote project`.
## Root cause
`agent_ui::ui::mention_crease::open_skill_file` was calling
`Project::create_local_buffer` to display the embedded read-only content
of a built-in skill:
```rust
let buffer = project.update(cx, |project, cx| {
project.create_local_buffer(content, None, false, cx)
});
```
`Project::create_local_buffer` has an explicit panic guard for remote
projects, so any user who clicked a built-in skill mention while in an
SSH or collab project would crash Zed.
## Fix
The buffer is purely a display surface for binary-embedded skill content
— it has no on-disk backing, isn't searchable, and never needs to be
tracked by the project's buffer store. Build it directly with
`language::Buffer::local(content, cx)` instead, matching the pattern
used elsewhere in `agent_ui` (`thread_view.rs`, `message_editor.rs`,
`inline_assistant.rs`, etc.). This works for both local and remote
projects.
## Repro (before fix)
1. Connect Zed to a remote (SSH) project.
2. In the agent panel, click on a built-in skill mention (e.g. one
referencing `create-skill`).
3. Zed aborts with SIGABRT.
After the fix, the mention opens a read-only editor with the embedded
markdown content, identical behavior to local projects.
Release Notes:
- Fixed a crash when clicking a built-in skill mention in the agent
panel while connected to a remote project
When `read_file` returns a file outline instead of full contents, the
result is now wrapped in a plain fenced code block rather than tagging
it with the file's path.
Previously, the outline was wrapped in a fenced block tagged with the
file path (e.g. `` ```crates/agent/src/tools/read_file_tool.rs ``).
Because the tag contains a slash, the markdown parser routed it through
`CodeBlockKind::FencedSrc`, resolved the file's language by path, and
ran the language's tree-sitter parser against the outline on every
paint. The outline is structural (e.g. `fn foo [L10-20]`), not actual
source for the file's language, so the parse was both expensive and
produced incorrect highlighting.
Because GPUI rebuilds the visible element tree on every window paint,
anything that triggers a repaint (cursor blink in the focused message
editor, animations, the turn timer, etc.) would re-run the tree-sitter
parse on the entire outline, throttling the frame rate while the tool
call was expanded.
This change adds an `is_outline_response` flag in `ReadFileTool::run`
and, when set, passes an empty tag to `MarkdownCodeBlock` so the
renderer sees `CodeBlockKind::Fenced` (no language). Plain monospace
formatting is preserved; the path tag is still used for the non-outline
(full file) case.
Adds two regression tests: one asserting the outline path uses an
untagged fenced block, and one asserting the full-file path keeps the
path tag (so the next person fixing this doesn't accidentally strip the
tag everywhere).
Also includes a small markdown-rendering follow-up:
`MarkdownElementBuilder::push_text` was recomputing the base text style
(cloning `base_text_style` and walking the style stack) twice per
highlighted token. For a code block with hundreds of highlight tokens,
that's hundreds of redundant `TextStyle` clones per paint. The style
stack does not change while runs are being attributed, so the style is
now computed once outside the loop and reused.
Closes AI-234
Release Notes:
- Improved scrolling smoothness in the agent panel when a `read_file`
tool call with a large file outline is expanded.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- N/A or Added/Fixed/Improved ...
