This adds ability to the assistant panel's context editor to accept
files being dropped on it.
Multiple things can be dropped on the assistant panel:
- project panel entries (one or many)
- tabs (one)
- external files (one or many)
Release Notes:
- N/A
Demo:
https://github.com/user-attachments/assets/fddee751-cbdf-4e2c-ac80-35dfb857cc8a
Co-authored-by: Bennet <bennet@zed.dev>
This PR adds the initial groundwork for invoking tools in response to
tool uses from the model.
Tool uses are run when the model responds with a `stop_reason` of
`tool_use`.
Currently the tool results are just inserted as text into the user
message. We'll want to include these as `tool_result` content on the
message, but Claude seems to understand it regardless.
Release Notes:
- N/A
This PR makes it so we propagate the `stop_reason` from Anthropic up to
the Assistant so that we can take action based on it.
The `extract_content_from_events` function was moved from `anthropic` to
the `anthropic` module in `language_model` since it is more useful if it
is able to name the `LanguageModelCompletionEvent` type, as otherwise
we'd need an additional layer of plumbing.
Release Notes:
- N/A
This adds a new button that on click, accepts the suggestion but does
not run the generated command.
https://github.com/user-attachments/assets/426b0ff3-8e19-435a-aa7f-89e062aefd4c
@danilo-leal @iamnbutler Any ideas on how to make both options
discoverable without having an extra button?
Release Notes:
- Added a way to accept terminal inline assist suggestions without
executing them
---------
Co-authored-by: Danilo Leal <67129314+danilo-leal@users.noreply.github.com>
This PR adds a tool registry to hold tools that can be called by the
Assistant.
Currently we just have a `now` tool for retrieving the current datetime.
This is all behind the `assistant-tool-use` feature flag which currently
needs to be explicitly opted-in to in order for the LLM to see the
tools.
Release Notes:
- N/A
This PR adjusts the approach we use to encoding tool uses in the
completion response to use a structured format rather than simply
injecting it into the response stream as text.
In #17170 we would encode the tool uses as XML and insert them as text.
This would require then re-parsing the tool uses out of the buffer in
order to use them.
The approach taken in this PR is to make `stream_completion` return a
stream of `LanguageModelCompletionEvent`s. Each of these events can be
either text, or a tool use.
A new `stream_completion_text` method has been added to `LanguageModel`
for scenarios where we only care about textual content (currently,
everywhere that isn't the Assistant context editor).
Release Notes:
- N/A
This is a refactor to prepare for adding LSP support in SSH remote
projects.
Release Notes:
- N/A
---------
Co-authored-by: Mikayla <mikayla@zed.dev>
Co-authored-by: Conrad <conrad@zed.dev>
This PR reverts #17173, as it introduced a segfault when opening any
Gleam project and the language server starts up.
This reverts commit a850731b0e.
Release Notes:
- N/A
Context: https://x.com/fasterthanlime/status/1819120238228570598
Up to this PR:
- We were not watching paths outside of a worktree when language server
requested it.
- We expected GlobPattern used for file watching to be always rooted at
the worktree root.
'1 mattered for observing global files (e.g. global RA config) and both
points had impact on "monorepos".
Let's picture the following scenario:
You're working on a Rust project that has two crates: bin and lib crate:
```
my-rust-project/
bin-crate/
lib-crate/
```
Up to this PR, making changes like changing field visibility in
lib-crate **was not reflected** in bin-crate until RA was restarted. RA
for bin-crate asked us to watch lib-crate. Now, depending on if you had
this project open as:
- a project with one worktree rooted at my-rust-project:
- due to '2, we never noticed that we have to notify RA instance for
bin-crate about changes in lib-crate.
- a project with two worktrees (bin-crate and lib-crate):
- due to '1 (as lib-crate is not within bin-crate's worktree), we once
again missed the fact that we have to watch for changes in lib-crate.
This PR solves this by introducing a side-channel - we just store fs
watchers for abs paths at the Project level. Worktree changes handling
is left relatively untouched - as it's used for other changes besides
LSP change notifying, I've figured to better leave it as is, as right
now we have 1 worktree change watcher; if we were to change it, we'd
have `(language server) + 1` watchers per worktree, which seems.. pretty
horrid.
What's the end effect? At the very least fasterthanlime should be a tad
happier; in reality though, I expect it to have some impact on LS
reliability in monorepo setups.
TODO
- [x] Wire through FileChangeType into `fs::watch` interface.
Release Notes:
- Improved language server reliability in multi-worktree projects and
monorepo. We now notify the language server more reliably about which
files have changed.
For ssh remoting lsps we'll need to have language server support
factored out of project.
Thus that begins
Release Notes:
- N/A
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Mikayla <mikayla@zed.dev>
This PR updates the Assistant with support for receiving tool uses from
Anthropic models and capturing them as text in the context editor.
This is just laying the foundation for tool use. We don't yet fulfill
the tool uses yet, or define any tools for the model to use.
Here's an example of what it looks like using the example `get_weather`
tool from the Anthropic docs:
<img width="644" alt="Screenshot 2024-08-30 at 1 51 13 PM"
src="https://github.com/user-attachments/assets/3614f953-0689-423c-8955-b146729ea638">
Release Notes:
- N/A
Context servers might return CR characters, which are not acceptable in
Zed and cause ranges to be invalidated. We need to normalize them.
Closes#17109
Release Notes:
- context_servers: Fixed an issue where context servers returning a
carriage return character would result in a panic.
This PR adds support for completions via MCP. The protocol now supports
a new request type "completion/complete"
that can either complete a resource URI template (which we currently
don't use in Zed), or a prompt argument.
We use this to add autocompletion to our context server slash commands!
https://github.com/user-attachments/assets/08c9cf04-cbeb-49a7-903f-5049fb3b3d9f
Release Notes:
- context_servers: Added support for argument completions for context
server prompts. These show up as regular completions to slash commands.
## Todo
* [x] Parse and present new XML output
* [x] Resolve new edits to buffers and anchor ranges
* [x] Surface resolution errors
* [x] Steps fail to resolve because language hasn't loaded yet
* [x] Treat empty `<symbol>` tag as None
* [x] duplicate assists when editing steps
* [x] step footer blocks can appear *below* the following message header
block
## Release Notes:
- N/A
---------
Co-authored-by: Mikayla <mikayla@zed.dev>
Co-authored-by: Peter <peter@zed.dev>
Co-authored-by: Marshall <marshall@zed.dev>
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Server can now include an optional description in a `prompts/get`
response. Zed will displayed the description as label of the slash
command.
Release Notes:
- context_servers: Servers can provide an optional description in
`prompts/get` responses that is displayed as the slash command label.
This PR fixes two issues:
1. The prompt library window didn't set an `app_id` on Linux, which
caused it to be missing the Zed logo
2. A dangling reference to the window in the Wayland client code, which
caused the prompt library window not to close. See:
https://github.com/zed-industries/zed/pull/13201
Release Notes:
- Linux: Fixed the prompt library not closing on Wayland
---------
Co-authored-by: Junkui Zhang <364772080@qq.com>
This adds a new action: `editor: toggle inline completions`.
It allows users to toggle inline completions on/off for the current
buffer.
That toggling is not persistent and when the editor is closed, it's
gone.
That makes it easy to disable inline completions for a single text
buffer, for example, even if you want them on for other buffers.
When toggling on/off, the toggling also overwrites any language
settings. So if you have inline completions disabled for Go buffers,
toggling them on takes precedence over those settings.
Release Notes:
- Added a new editor action to allow toggling inline completions
(Copilot, Supermaven) on and off for the current buffer, taking
precedence over any settings.
Co-authored-by: Antonio <antonio@zed.dev>
We've noticed performance issues in long conversations with assistants;
the profiles pointed to slowiness in WrapMap (and indeed there were some
low hanging fruits that we picked up in
https://github.com/zed-industries/zed/pull/16761). That however did not
fully resolve the issue, as WrapMap still cracked through in profiles;
basically, the speedup I've landed has just moved the post elsewhere.
The higher level issue is that we were trying to refresh message headers
for all messages, irrespective of whether they've actually needed to be
updated. This PR fixes that by using `replace_blocks` API where
possible.
Release Notes:
- Improved performance of Assistant Panel with long conversations.
Closes#16804
Similar to #15708, when reading prompts from a template, both Windows
and Linux might end up with `CRLF (\r\n)` line endings, which can result
in a panic.
Release Notes:
- N/A
* Linux Clippy lints fixed
* Zed local tasks are now simpler to rerun
* Zed's `release-fast` build profile keeps the debug info so it's
possible to properly debug things without altering the sources
Release Notes:
- N/A
Follow-up https://github.com/zed-industries/zed/pull/16085 that fixes
the search deploy to be actually a part of the terminal-related
bindings.
Part of https://github.com/zed-industries/zed/issues/16839
Also
* fixes few other bindings to use `shift` and avoid conflicts with the
existing key bindings.
* adds terminal inline assist to the context menu and makes both the
menu and the button to dynamically adjust to `assist.enabled` settings
change
It is still unclear to me, why certain labels for certain bindings are
wrong (it's still showing `ctrl-w` for closing the terminal tab, and
`shift-insert` instead of `ctrl-shift-v` for Paste, while Insert is near
and has a `ctrl-shift-c` binding shown) but at least the keys work now.
Release notes:
- Improved Linux terminal keymap and context menu
### Pull Request Title
Introduce `max_output_tokens` Field for OpenAI Models
https://platform.deepseek.com/api-docs/news/news0725/#4-8k-max_tokens-betarelease-longer-possibilities
### Description
This commit introduces a new field `max_output_tokens` to the OpenAI
models, which allows specifying the maximum number of tokens that can be
generated in the output. This field is now integrated into the request
handling across multiple crates, ensuring that the output token limit is
respected during language model completions.
Changes include:
- Adding `max_output_tokens` to the `Custom` variant of the
`open_ai::Model` enum.
- Updating the `into_open_ai` method in `LanguageModelRequest` to accept
and use `max_output_tokens`.
- Modifying the `OpenAiLanguageModel` and `CloudLanguageModel`
implementations to pass `max_output_tokens` when converting requests.
- Ensuring that the `max_output_tokens` field is correctly serialized
and deserialized in relevant structures.
This enhancement provides more control over the output length of OpenAI
model responses, improving the flexibility and accuracy of language
model interactions.
### Changes
- Added `max_output_tokens` to the `Custom` variant of the
`open_ai::Model` enum.
- Updated the `into_open_ai` method in `LanguageModelRequest` to accept
and use `max_output_tokens`.
- Modified the `OpenAiLanguageModel` and `CloudLanguageModel`
implementations to pass `max_output_tokens` when converting requests.
- Ensured that the `max_output_tokens` field is correctly serialized and
deserialized in relevant structures.
### Related Issue
https://github.com/zed-industries/zed/pull/16358
### Screenshots / Media
N/A
### Checklist
- [x] Code compiles correctly.
- [x] All tests pass.
- [ ] Documentation has been updated accordingly.
- [ ] Additional tests have been added to cover new functionality.
- [ ] Relevant documentation has been updated or added.
### Release Notes
- Added `max_output_tokens` field to OpenAI models for controlling
output token length.
This PR allows configuring the intensity of code fade applied to unused
code (relates to #4785).
_Note: Right now I included it as a top level config which might be a
little out of place (but is easiest to instrument). Open for suggestions
on where else it would be better suited for._
_Note 2: I am unfamiliar with the codebase. Feel free to either close
this PR and re-implement in a better way or suggest high level changes
if I'm approaching anything wrong :)._
Release Notes:
- Added `unnecesary_code_fade` setting to control how strongly to fade
unused code.
---------
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
Follow-up to https://github.com/zed-industries/zed/pull/16471
* Don't expand the default prompt by default, since it looks strange in
the expanded state
* Don't create two `You` headers by default. Just insert a blank line
after the default prompt.
Release Notes:
- N/A
Release Notes:
- Adds UI affordances to the assistant panel to show which messages have
been cached
- Migrate cache invalidation to be based on `has_edits_since_in_range`
to be smarter and more selective about when to invalidate the cache and
when to fetch.
<img width="310" alt="Screenshot 2024-08-16 at 11 19 23 PM"
src="https://github.com/user-attachments/assets/4ee2d111-2f55-4b0e-b944-50c4f78afc42">
<img width="580" alt="Screenshot 2024-08-18 at 10 05 16 PM"
src="https://github.com/user-attachments/assets/17630a60-7b78-421c-ae39-425246638a12">
I had originally added the lightening bolt on every message and only
added the tooltip warning about editing prior messages on the first
anchor, but thought it looked too busy, so I settled on just annotating
the last anchor.
This fixes a weird bug:
1. Use `/workflow` in assistant
2. Have it generate a step that modifies a file
3. Either (a) select the step in the assistant and have it auto-insert
newlines (b) select "Transform" to have the step applied
4. Close the modified file in the editor ("Discard")
5. Re-open the file
6. BUG: the changes made by assistant are still there!
The reason for the bug is that the assistant keeps references to buffers
and they're not closed/reloaded when closed/reopened.
To fix the bug we now rollback the applied workflow steps when
discarding a buffer.
(This does *not* yet fix the issue where a workflow step inserts a new
buffer into the project/worktree that does not show up on the file
system yet but in `/file` and hangs around until Zed is closed.)
Release Notes:
- N/A
Co-authored-by: Bennet <bennet@zed.dev>
This PR adds an extra item to the slash command picker that links users to the doc that teaches how to create a custom one.
Release Notes:
- N/A
---------
Co-authored-by: Danilo Leal <67129314+danilo-leal@users.noreply.github.com>
