Commit graph

11 commits

Author SHA1 Message Date
Agus Zubiaga
5615be51cc
assistant edit tool: Revert fuzzy matching (#26996)
#26935 is leading to bad edits, so let's revert it for now. I'll bring
back a version of this, but it'll likely just focus on indentation
instead of making the whole search fuzzy.

Release Notes: 

- N/A
2025-03-18 13:08:09 -03:00
Agus Zubiaga
94b63808e0
assistant edit tool: Fuzzy match search block (#26935)
Release Notes:

- N/A

Co-authored-by: Antonio Scandurra <me@as-cii.com>
2025-03-17 18:33:20 +00:00
Agus Zubiaga
1f62274a89
assistant edit tool: Return applied actions back to main model (#26810)
We'll now include the search/replace block that got applied as part of
the tool output. We think this will help the model have a better idea of
how the file changed and prevent later edit failures.

Release Notes:

- N/A
2025-03-17 13:21:35 -03:00
张小白
0b492c11de
Use line_endings macro for the edit tool tests (#26642)
This aligns with how we handle other tests on Windows.

Release Notes:

- N/A
2025-03-15 14:16:10 +08:00
Antonio Scandurra
70c973f6c3
Fix issues in EditFilesTool, ListDirectoryTool and BashTool (#26647)
Release Notes:

- N/A
2025-03-13 09:41:27 +00:00
Agus Zubiaga
606aa7a78c
Edit tool debugging (#26637)
Adds an `debug: edit tool` action that opens a new view which will help
us debug the edit tool internals. As the edit tool runs, the log
displays:

- Instructions provided by the main model
- Response stream from the editor model
- Parsed edit blocks
- Tool output provided back to main model

The log automatically records all edit tool interactions for staff, so
if you notice something weird, you can debug it retroactively without
having to open the debug tool first. We may want to limit the number of
recorded requests later.

I have a few more ideas for it, but this seems like a good starting
point.


https://github.com/user-attachments/assets/c61f5ce8-08b1-4500-accb-db2a480eb3ab


Release Notes:

- N/A
2025-03-13 04:03:01 +00:00
Agus Zubiaga
5f74297576
Fix edit tool tests on windows (#26552)
Assertions on the parsed system prompt should use CRLF on Windows. I
didn't see it before because I was testing on my Windows VM from a
shared folder I cloned on macOS.

Release Notes:

- N/A
2025-03-12 15:52:51 +00:00
Antonio Scandurra
41eb586ec8
Remove list_worktrees and use relative paths instead (#26546)
Release Notes:

- N/A
2025-03-12 15:06:04 +00:00
Agus Zubiaga
669c6a3d5e
assistant edit tool: Do not include \r in old/new str (#26542)
#26538 fixed part of the issue, but it would keep trailing carriage
returns in the old/new strings. The model is unlikely to produce those,
but we might as well support them.

Release Notes:

- N/A
2025-03-12 11:34:40 -03:00
Agus Zubiaga
6b56fee6b0
assistant edit tool: Support \r\n around markers (#26538)
This should fix the tests on Windows

Release Notes:

- N/A
2025-03-12 11:00:16 -03:00
Agus Zubiaga
47a89ad243
assistant: Edit files tool (#26506)
Exposes a new "edit files" tool that the model can use to apply
modifications to files in the project. The main model provides
instructions and the tool uses a separate "editor" model (Claude 3.5 by
default) to generate search/replace blocks like Aider does:

````markdown
mathweb/flask/app.py
```python
<<<<<<< SEARCH
from flask import Flask
=======
import math
from flask import Flask
>>>>>>> REPLACE
```
````

The search/replace blocks are parsed and applied as they stream in. If a
block fails to parse, the tool will apply the other edits and report an
error pointing to the part of the input where it occurred. This should
allow the model to fix it.


Release Notes:

- N/A

---------

Co-authored-by: Antonio Scandurra <me@as-cii.com>
2025-03-12 12:30:47 +00:00