Refs #29733
This pull request introduces a new field to the `StreamingEditFileTool`
that lets the model create or overwrite a file in a streaming way. When
one of the `assistant.stream_edits` setting / `agent-stream-edits`
feature flag is enabled, we are going to disable the `CreateFileTool` so
that the agent model can only use `StreamingEditFileTool` for file
creation.
Release Notes:
- N/A
---------
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
Co-authored-by: Oleksiy Syvokon <oleksiy.syvokon@gmail.com>
This PR prevents any unnecessary lines from being rendered in the edit
file tool card in the case of small diffs.
I think this (hopefully) addresses the last remaining task from
https://github.com/zed-industries/zed/pull/29448.
| `main` | This PR |
| --- | --- |
| <img width="634" alt="main"
src="https://github.com/user-attachments/assets/7c06394e-957a-4d36-a484-5974687041e9"
/> | <img width="634" alt="PR"
src="https://github.com/user-attachments/assets/84206d5a-a93a-4a42-99ca-7cdebb0d91bb"
/> |
(The last empty line in the second image is an empty line present in the
file itself)
---
n the second commit I also preemtively disabled vertical overscrolling
for full mode editors which are sized by content. This is basically the
same fix as in https://github.com/zed-industries/zed/pull/28471.
Strictly speaking, this is not needed for the fix here, but I thought it
might be nice to have for the future to prevent any issues from occuring
due to overscroll.
Release Notes:
- agent: Improved rendering of small diffs for the edit file tool card.
This PR updates the max mode toggle to use the muted color.
This makes it fit in more with the rest of the controls.
<img width="243" alt="Screenshot 2025-05-01 at 5 24 01 PM"
src="https://github.com/user-attachments/assets/57267d29-3c7b-4ea9-b6b9-81c42f6b7e1c"
/>
Release Notes:
- agent: Adjusted the color of the max mode toggle.
Release Notes:
- Fixed a bug where clicking context pills switched into the "editing
message" state instead of clicking the pill.
Co-authored-by: Michael <michael@zed.dev>
Co-authored-by: Ben <ben@zed.dev>
Closes#29753
The template contains an error: `has_default_user_rules` is always
undefined and should be `has_user_rules` instead.
Release Notes:
- Fixed default user rules ignored during prompt building.
We weren't using this one anymore. We used to use it for the switch that
toggled tools on, which doesn't exist anymore.
Release Notes:
- N/A
---------
Co-authored-by: Joseph T. Lyons <josephtlyons@gmail.com>
This pull request introduces a new tool for streaming edits. The
short-term goal is for this tool to replace the existing `EditFileTool`,
but we want to get this out the door as soon as possible so that we can
start testing it.
`StreamingEditFileTool` is mutually exclusive with `EditFileTool`. It
will be enabled by default for anyone who has the `agent-stream-edits`
feature flag, as well as people that set `assistant.stream_edits` to
`true` in their settings.
### Implementation
Streaming is achieved by requesting a completion while the `edit_file`
tool gets called. We invoke the model by taking the existing
conversation with the agent and appending a prompt specifically tailored
for editing. In that prompt, we ask the model to produce a stream of
`<old_text>`/`<new_text>` tags. As the model streams text in, we
incrementally parse it and start editing as soon as we can.
### Evals
Note that, as part of this pull request, I also defined some new evals
that I used to drive the behavior of the recursive LLM call. To run
them, use this command:
```bash
cargo test --package=assistant_tools --features eval -- eval_extract_handle_command_output
```
Or comment out the `#[cfg_attr(not(feature = "eval"), ignore)]` macro.
I recommend running them one at a time, because right now we don't
really have a way of orchestrating of all these evals. I think we should
invest into that effort once the new agent panel goes live.
Release Notes:
- N/A
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Co-authored-by: Oleksiy Syvokon <oleksiy.syvokon@gmail.com>
Release Notes:
- agent: Support adding/removing context when editing existing message
---------
Co-authored-by: Cole Miller <m@cole-miller.net>
Co-authored-by: Cole Miller <cole@zed.dev>
We have a report of a panic when indexing into
`BufferConflicts.block_ids` using the `old_range` from the
`ConflictsUpdated` event, indicating that the `block_ids` array can get
out of sync with the underlying `ConflictSet`. This PR adds a mitigation
so that we won't panic in this situation, as a stopgap until the bug can
be reproduced in a test and fixed at the root.
Release Notes:
- N/A
⚠️ Work in progress until all of the to-dos are knocked out:
- [x] Disable soft-wrapping
- [x] Make it foldable only after a certain number of lines
- [x] Display tool status errors
- [x] Fix horizontal scroll now that we've disabled soft-wrap
- [ ] Don't render unnecessary extra lines (will be added later, on a
follow-up PR)
Release Notes:
- N/A
---------
Co-authored-by: Agus Zubiaga <hi@aguz.me>
Co-authored-by: Michael Sloan <mgsloan@gmail.com>
This PR updates the git store to not register a change in a repository's
merge heads until conflicted paths are seen.
We currently use the repository's merge heads only to decide when the
list of conflicted paths should be refreshed. Previously, the logic
looked like this:
- Whenever we see a change in the merge heads, set the list of
conflicted paths by filtering the output of `git status`.
It turns out that when a conflicting merge takes a while, we can see
this sequence of events:
1. We get an event in .git and reload statuses and merge heads.
Previously there were no merge heads, and now we have some, but git
hasn't finished figuring out which paths have conflicts, so we set the
list of conflicted paths to `[]`.
2. Git finishes computing the list of conflicted paths, and we run
another scan that picks these up from `git status`, but then we throw
them away because the merge heads are the same as in (1).
By not updating our stored merge heads until we see some conflicts in
`git status`, we delay this step until (2), and so the conflicted paths
show up in the git panel as intended.
This means that our merge heads state no longer matches what's on disk
(in particular, during a clean merge we'll never update them at all),
but that's okay because we only keep this state for the purpose of
organizing conflicts.
Release Notes:
- Fixed a bug that could cause conflicted paths to not appear in their
own section in the git panel.
Maybe "extra" isn't the best word, but I'm referring to the ellipsis
menu on the far right of the panel that holds "extra" or "additional
options". There is a known issue with context menus, at least if
implemented this way, that makes hitting enter to select an option not
work. Will leave this fix to a later PR.
Release Notes:
- N/A
This new default profile is one that doesn't use any tools; it's
completely "naked" and it shouldn't lean into trying to read things from
the current project at hand. Better suited for general topic chats with
the LLM.
PS: Still expecting some wordsmithing here before merging.
Release Notes:
- agent: Added a new default profile called "Manual" that doesn't
include any tools, for general topic chats with the LLM.
This reverts commit e661a0afd6.
Closes #ISSUE
Release Notes:
- Reverted changes to Python subroot detection which could have caused
multiple python processes to be spawned when working in projects with
multiple `pyproject.toml` files.
DebugSession.mode is no longer needed because project::debugger::Session
manages its own state now (booting, running, terminated), and removing
mode simplifies a lot of the code that uses running state.
I used Zed AI to do a good chunk of the refactor, but I doubled-checked
everything it did and changed a good amount of its updates.
Release Notes:
- N/A
Co-authored-by: Zed AI <ai@zed.dev>
Fixed issue where eval thread judges were not considering the last
response in the thread.
The problem was that they were getting the full list of messages from
`last_request`, which (being a request!) did not have the response yet.
Release Notes:
- N/A
- [x] Ensure what appears in the dropdown is really what is accurate
- [x] Ensure keyboard navigation works:
- [x] Switching tabs with `enter`
- [x] Closing items from the menu item
- [x] Opening the dropdown
- [x] Focus assistant panel on dismiss
- [x] Add ability to close items from the dropdown menu
- [x] Persistence
- [x] Correct behavior when opening a text thread
Release Notes:
- agent: Added a navigation menu that shows the recently opened threads.
The button to see the full history view has been changed inside this
menu.
---------
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Co-authored-by: Cole Miller <m@cole-miller.net>
Co-authored-by: Bennet Bo Fenner <bennet@zed.dev>
Co-authored-by: Cole Miller <cole@zed.dev>
Adds popular examples of long-running commands to system prompt.
Unfortunately, I couldn't add an eval example as the new terminal tool
no longer works in `eval`. We can look into that tomorrow, but I'm
seeing improvements when manually testing this, so I'd like to merge it.
<img
src="https://github.com/user-attachments/assets/ac24e617-e068-466f-875d-c30e1f2465c4"
width=400></img>
Release Notes:
- agent: Discourage long-running commands
## Why?
Some users expressed a preference for the AnyQuotes and AnyBrackets text
objects to align more closely with traditional Vim behavior, rather than
the mini.ai plugin's approach. To address this, I’ve introduced two new
text objects: MiniQuotes and MiniBrackets. These retain the mini.ai
plugin behavior, while the updated AnyQuotes and AnyBrackets now follow
the logic described in [this bug
report](https://github.com/zed-industries/zed/issues/25563) and [this
bug report](https://github.com/zed-industries/zed/issues/25562).
## Behavior Overview:
### AnyQuotes and AnyBrackets:
These now prioritize the innermost range first (e.g., the closest quotes
or brackets). If none are found, they fall back to searching the current
line. This aligns with the behavior requested in the issue.
### MiniQuotes and MiniBrackets:
These maintain the mini.ai plugin behavior, prioritizing the current
line before expanding the search outward.
### Usage Examples:
AnyQuotes: Works like ```ci', ci", ci` , ca', ca", ca` , etc.```
AnyBrackets: Works like ```ci(, ci[, ci{, ci<, ca(, ca[, ca{, ca<,
etc.```
Please give these changes a try and let me know your thoughts!
### Release Notes:
- vim: Add AnyQuotes, AnyBrackets, MiniQuotes and MiniBrackets text
objects
---------
Co-authored-by: Ben Kunkle <ben@zed.dev>
