Over the past releases, many users have noticed a growing number of bugs, UI inconsistencies, and general UX issues. While we appreciate the new features, the constant stream of small “papercuts” makes the app frustrating to use. As an Early Access user, I expect some issues, but the experience is starting to feel more like an alpha test, especially when major regressions occur (like these 5 recent issues of the terminal).
I believe it’s time to propose that the next major version be a dedicated “Bug Catching Update”.
Instead of adding new functionality, please focus one development cycle entirely on:
Fixing long-standing bugs.
Addressing technical debt.
Polishing the overall user experience and stability.
A stable and reliable application is the best feature of all. It would make the existing user base incredibly happy and create a stronger foundation for future development.
If you agree, please vote for and share this suggestion.
Have to add my backing here. We are already seeing 1.7.x versions, and the totally broken terminal issue (terminals from the agent never execute at all anymore), which first appeared after 1.6.27, is still not fixed. I just tried 1.7.11 today, and the terminal issue is STILL present. I simply cannot use Cursor at all, if the agent cannot run terminal commands. Nearly everything I do involves terminal commands, outside of pure research (whcih is also largely borked due to broken MCP support, broken @Docs, support, broken or simply unavailable @Web suport in many models, etc.)
Last time Cursor upgraded the VSCode version, it was the best update yet—seemingly inspiring the engineering team to thoroughly refactor their code. But now, as they continue making updates on top of it, it feels like a recent graduate starting to mess around with the codebase: every new feature they add comes bundled with 10+ bugs.
Looking forward to upgrading the VSCode version soon.
Performance has severely degraded when managing multiple parallel chats. I have a fairly beefy Windows PC, but the app crawls to a freeze despite the CPU/Memory resources being around 30% utilization.
Latest update causes chats to completely break after certain Terminal commands (usually related to lint checks), I have to continue the chat in a new chat the references the old one.
if it won’t let you interact with sudo commands to type in your password set up an alias for pkexec or whatever it is called. It lets you have cursor bring up the login prompt for most ubuntu based distros.
I’ve been facing serious problems with VibeCoding (Cursor) while developing my Laravel + Flutter project.
Too many errors – The system frequently throws unrelated or repetitive errors, even in working code.
Fixes take too long – It often takes 1–2 days to resolve even simple issues.
Changes don’t apply – The updates or fixes I make do not reflect in the project files or runtime environment.
High token usage – Because the same operations repeat and fail, token consumption is extremely high.
Project delays – These issues are extending the project completion time significantly.
Reinstalling Cursor doesn’t help – Even after deleting memory, reinstalling Cursor, and clearing all caches, the same problems continue.
Summary:
VibeCoding (Cursor) struggles to properly apply code changes, keeps reintroducing old errors, consumes a lot of tokens, and causes major development delays. I need a stable environment where updates reflect immediately and error fixing is consistent.
Would you like me to make this version sound more professional and polite, as if you’re writing a support ticket or official complaint to Cursor?
Problem Summary – VibeCoding / Cursor Environment Issues
I’m currently using Cursor (VibeCoding) with GPT-4.5 and GPT-5 models, on a $200/month plan, and I also have a ChatGPT Plus subscription.
Despite the high-tier setup, I’m facing major development issues that are severely slowing down my project.
Issues Observed
Frequent and repetitive errors – Cursor often throws unrelated or random logic errors, even when the code is valid.
Slow and unreliable fixes – Error resolution sometimes takes days, and the system reintroduces the same issues later.
Changes not reflecting – Code updates or fixes I apply don’t consistently appear in the project; it often reverts or ignores edits.
Excessive token consumption – Because of repeated reprocessing and failed attempts, token usage is extremely high.
Extended project timelines – These issues have caused significant development delays.
No improvement after full reset – Even after deleting caches, clearing all memory, and reinstalling Cursor, the same issues persist.
Summary
I’m investing in premium AI tools (Cursor + ChatGPT Plus with GPT-5 access) to accelerate my Laravel + Flutter project,
but currently, VibeCoding (Cursor) fails to apply updates properly, frequently breaks working code, consumes large amounts of tokens, and causes major time loss.
I need a stable and synchronized environment where updates immediately reflect, caching issues are fixed, and GPT models respond consistently.
Act as a software engineer.
Use tool cursor_agent_chat with attribute new_chat = true, mode = Agent and model = gpt-5-high.
Also, past text below in chat_message_field and click send_message_to_agent button.
- Analyze my project and make an analytical report on how it meets modern software development guidelines.
- Give recommendations on how to improve the situation.
Bring the report right in the chat