As part of our efforts to improve the agent harness in Cursor 2.0, we've significantly improved the quality of GPT-5-Codex responses.
It can now work uninterrupted for much longer, with reduced overthinking and more accurate edits. Enjoy!
Finally, we've drastically improved the performance of using LSPs.
Python and TypeScript LSPs are now faster by default. Memory is dynamically configured based on available RAM.
We've also fixed a number of memory leaks and improved memory usage.
Give me the option to let it execute its own plan autonomously without having to reprompt it just to continue what it already knows it needs to do every 60 seconds—it's really annoying.
Курсор, Кодекс и Копилот - это костыли для тех, кто пишет по инструкции, не чувствуя живой логики систем. Они помогают воспроизводить старые паттерны, но не создают новые формы.
these are huuuuuge!
BTW, there are some bugs in the plan preview & markdown view.
I couldnt search using CMD + F, which is kinda annoying + it auto scrolls up/down when i switch files & come back.
Am I the only one confused about which files are passed in context and which aren’t? Previous UI was much more transparent, the new one adds some uncertainty.
🔥 Loving these updates, especially the improved LSP performance and the new background agent UI. Cursor just keeps getting smoother every release. 🚀
Can’t wait to test the cloud agents and new review flow in action! 💻✨
When I click the "Add New MCP Service" button, it opens the previously installed MCP, and I cannot add a new MCP. This issue still hasn't been improved.