We were including the context at the end which meant it never got
cached. We'll now include it with the first message that introduced it
so it's cached as long as it doesn't change.
This is an improvement, but we probably still need to think of ways to
optimize caching for cases where files in context change.
Release Notes:
- N/A
It's super easy to undo those changes. In a future PR, we should also
avoid requiring confirmation in the batch tool if all the underlying
tools don't require confirmation.
Release Notes:
- N/A
This PR removes an unneeded `else if` branch that was the same as the
previous branch in the conditional.
Doesn't seem necessary to run this twice.
Release Notes:
- N/A
This PR makes it so if we receive a tool use from the model without any
corresponding text, we'll insert some placeholder text to keep the
conversation in a well-structured format.
This aims to fix an error that Antonio was seeing:
Release Notes:
- N/A
This seems to improve the performance of `gemini-2.5-pro-exp-03-25`
significantly.
We know define a single `Tool` that has multiple `FunctionDeclaration`s,
instead of defining multiple `Tool`s with a single
`FunctionDeclaration`.
Oddly enough the `flash` models seemed to work perfectly fine with the
multiple `Tool { ... }` definitions
Release Notes:
- N/A
- bump our livekit version to include a fix for a crane bug (TODO: add
link when an issue is filed on crane)
- switch to a clang stdenv for both linux and macos
- manually unify versions of our notify crate
- remove old linker flags which were only needed for livekit
- fix an issue where RUSTFLAGS shadowed the rustflags from cargo configs
Release Notes:
- N/A
We'll now prompt the user to start a new thread when the active one gets
too long.
<img width=500
src="https://github.com/user-attachments/assets/91445bc0-3e81-422f-aa4a-b8f0741f9d9a"></img>
When they click "Start New Thread", will create a new one with the
previous one added as context.
<img width=500
src="https://github.com/user-attachments/assets/c3b4223f-5bdd-4ba4-956f-5a5880d5e2c3"></img>
Instead of including the full thread text, we'll now add summarized
versions of threads to the context, allowing you to continue the
conversation even if it was near the token limit.
- Thread summaries are cached and persisted.
- A cached summary is invalidated if the thread is continued.
- We start generating the thread summary as soon as it's selected from
the picker. Most times, the summary will be ready by the time the user
sends the message.
- If the summary isn't ready by the time a message is sent, the user
message will be displayed in the thread immediately, and a "Summarizing
context..." indicator will appear. After the summaries are ready, we'll
start generating the response and show the usual "Generating..."
indicator.
Release Notes:
- N/A
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
Closes#20391
### Summary
This adds a persistence layer to the command palette so that usages can
persist after Zed is closed and re-opened.
The current "usage" algorithm is unchanged, e.g.:
- Sorts by number of usages descending (no recency preference)
- Once a user's query is active, removes these suggestions in favor of
fuzzy matching
There are some additional considerations in order to keep the DB from
growing uncontrollably (and to make long-term use ergonomic):
- The "invocations" count handles max values (though at u16, it seems
unlikely a user will deal with this)
- If a command is un-invoked for more than a month, it stops being
considered a recent usage, and its next update will update its usages
back to 1
### Future Considerations
- Could make the "command expiry" configurable in settings, so the user
can decide how long to hold onto recent usages
- Could make a more sophisticated algorithm which balances recency and
total invocations - e.g. if I've used COMMAND_A 100 times in the last
month, but COMMAND_B 10 times today, should COMMAND_B actually be
preferred?
- Could do preferential fuzzy-matching against these matches once the
user starts a query.
Release Notes:
- Added persistent history of command palette usages.
---------
Co-authored-by: Peter Finn <mastion11@gmail.com>
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
This PR completes the process of moving git repository state storage and
scanning logic from the worktree crate to `project::git_store`.
Release Notes:
- N/A
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Conrad <conrad@zed.dev>
This PR renames the `assistant2` actions to `agent`.
Note that any `assistant` actions have been left as-is for now so that
there aren't any changes to users not in the feature flag.
Release Notes:
- N/A
This PR adds some affordances for when the currently-selected model does
not support tools.
We disable the profile selector and put it into a "No Tools" state:
<img width="1394" alt="Screenshot 2025-04-01 at 3 58 00 PM"
src="https://github.com/user-attachments/assets/de6ecb0f-7657-4e16-9d5d-7bbfbc2b0a5c"
/>
We will also only attach tools to the request to the model if the model
supports it.
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/27619
Fixes issue with right wrapped movement when a multi-byte character is
at the end of the line. This is done by grabbing the last character on
the current row and using that characters size to calculate the
`max_column` variable, which is used to decide if the next right
movement should move down the line or not.
We did notice a bit of code that could be an issue that we wanted to
call out.
[Here](https://github.com/zed-industries/zed/blob/main/crates/editor/src/display_map.rs#L1070)
inside of `clip_at_line_end` it also does a saturating_sub(1), assuming
a single byte character. We didn't run into any issues due to this line
but felt like a similar bug. We can apply a similar fix if wanted to
pose the question first.
Test case: Moving to next line when eol is a multi-byte character
https://github.com/user-attachments/assets/1021ab1f-f49d-4986-8f9a-8cfc7e5c91bc
Release Notes:
- Fixed issue in vim forward spacing when a multi-byte character is at
the eol
---------
Co-authored-by: KyleBarton <kjbarton4@gmail.com>
Python envs created with virtualenvwrapper have the ability to link to
the project directory (via the `.project` file in the virtualenv).
`python-environment-tools` supports this and reports the project path.
Additionally, some tools (e.g virtualfish) recognize special
"environment activation" files in project directories (`.venv` by
default)
[1].
Use the above information to sort reported Python toolchains so that the
correct env for a given worktree is at the top.
[1]
https://virtualfish.readthedocs.io/en/latest/plugins.html#auto-activation-auto-activation
Release Notes:
- python: Improved detection of virtualenvwrapper environments in work
trees
This PR updates the Agent Panel tooltip in the status to use "Agent
Panel" instead of "Assistant Panel".
Also changes the name we use in workspace serialization.
Release Notes:
- N/A
This PR makes it so we include the `default_profile` and `profiles`
settings in the JSON schema.
This provides completions when editing the `settings.json`.
Release Notes:
- N/A
This unblocks work on new debugger UI, where we don't want the
set_active function to be called unconditionally.
Release Notes:
- N/A
---------
Co-authored-by: Anthony Eid <hello@anthonyeid.me>
This PR adds a new `enable_all_context_servers` field to agent profiles
to allow them to enable all context servers without having to opt into
them individually.
The "Write" profile will now have all context servers enabled out of the
box.
Release Notes:
- N/A
Eager and Subtle modes are only supported for Zed as an edit prediction
provider and they were visible if your provider is Copilot, which is
misleading.
Release Notes:
- N/A
