Tom Ross
Tom Ross
What: - We currently support JS/TS/Python/Go for advanced documentation range and insertion point determination. How: - Add tree-sitter queries for more languages, use the most popular languages by completions suggested...
What: - We think we can improve the performance of “Ask Cody to Fix” by including nearby identifiers as additional context. How: - Use `bfg/contextForIdentifiers`, provide identifiers from the range...
`Browserslist` recommend that you update `caniuse-lite` at regularly through the following command: `npx browserslist@latest --update-db` We should be able to automate this through Renovate so we don't forget. Further context:...
## Description The current smart apply implementation uses an OS-native dropdown for "Insert at Cursor" and "Save to New File". This is via a VS Code API. It works well...
## Description Adds support for inline completion context WIP ## Test plan Manual testing
## Test plan
## Description **Image:** Switches from the custom diff implementation to just showing a unified diff image in the debug panel. 1. Fixes bug where lines were ordered correctly 2. Means...
## Description This PR adds support for client configured font-size/line height/pixel ratios for image generated suggestions. ## Test plan Covered by integration agent tests, see the screenshots generated in the...
## Description Using 3.7 Sonnet as a reasoning model breaks for inline edit because we prefix the response with text that isn't We should support reasoning models for inline edit,...
## Description This PR implements an in-memory cache for autoedit suggestion images using the `AutoeditRequestID` as the cache key. The cache prevents regenerating the same image for repeated requests, improving...