I have been an active user of your platform, previously using v0.dev extensively in recent months. I was extremely satisfied with the performance — the tool was able to take my PHP files and make precise, customized modifications. It generated visually appealing designs, implemented requested features flawlessly, and allowed me to build fully functional themed websites from scratch.
Back then, even if the code to be generated was over 3,000 lines, the platform would complete it successfully as long as sufficient credits were available. That level of performance made V0 an invaluable part of my workflow.
However, upon returning to use your service today, I noticed the platform has transitioned to a new domain, v0.app, and introduced a feature called Agent mode. Since this change, the tool has become significantly less effective and no longer meets my development needs.
Specifically:
I am now unable to have my existing PHP files updated and redesigned as before. Previously, I could submit a file and request updates in design and functionality, which V0 handled flawlessly. Now, none of these requests can be fulfilled.
My current PHP file consists of 1,522 lines, and I request improvements in both design and functionality. However, the process now consistently stops around line 600, and the request fails to complete. This issue only started occurring after the introduction of Agent mode.
Additionally, when I try to generate a complete themed site from scratch — something V0 previously did exceptionally well — the resulting themes are now very poor in both design and implementation.
The previous version of V0 was incredibly effective and productive for my work, and I was truly impressed by its capabilities. Unfortunately, the recent changes have made the platform almost unusable for my development workflow.
I would also like to know if these issues are being addressed, and if so, when a resolution can be expected. Knowing whether there are plans to restore or improve the functionality affected by Agent mode would be greatly appreciated, as it will influence my decision to continue using the platform.
Agent ErrorAn error occurred while the agent was working, Missing query: user_read_only_context/text_attachments/index-v0-son-hn298.php
I am providing my index.php file and requesting several updates. However, due to the current Agent Mode, the process fails.
When writing the following SVG code: <svg class="w-5 h-5 text
the generation stops and becomes corrupted. In v0 chat, it states that it successfully completed all of my requests, yet the code—which should exceed 1,522 lines—always stops at line 663. This occurs precisely when it reaches the SVG section, after which it ends prematurely.
For the past three days, it has failed at exactly the same line, producing no further progress. This Agent Mode has completely broken the functionality.
Jacob, could you please comment on when these issues will be resolved? At present, I am unable to use your platform. If we cannot receive proper support or responses, then what is the purpose of us opening threads here?
I’ve collected the problematic block info and passed it along to the v0 agent team – the stream does successfully complete but there’s an error writing to the file
In the meantime, whenever the LLM is messing up consistently I revert to a version before their mistake and then fork into a new chat so there’s no memory of the mistake
I read that you mentioned transitioning to a new editor. Does this mean that all the current issues will be resolved, and when will this transition take place?
A few hours ago, I was able to get a complete file output by starting a new chat, strengthening my prompt as you suggested, and repeatedly refining it. However, in order to reach that result, I had to write numerous prompts and start multiple new chats. I tested this strengthened prompt across many new conversations, but most of the time my .php file ended up broken, causing my website to become inaccessible. I am certain there were many errors in the code, although I did not examine all of them in detail. These problems began after the introduction of the V0 Agent Mode.
Previously, with V0 before Agent Mode, I could request any modification, and it would be completed in a single attempt—perfectly understanding my requirements. Now, even when using a strengthened prompt with Agent Mode enabled, two outcomes are common: either it produces faulty PHP code that breaks the file, or it delivers something closer to my request but with significant issues—such as numerous design errors in the front-end, unwanted additional elements, or incomplete implementation of my instructions.
Before Agent Mode, I never encountered these issues. I could achieve exactly what I wanted with one prompt, and if I didn’t like the front-end design, I would simply spend additional credits to have a new one created. Every front-end design produced back then looked flawless, and within two or three attempts, I could get something excellent. The back-end also worked perfectly—delivering both the front-end and back-end together in a single prompt with an ideal result. Now, because of Agent Mode, even after dozens of new chats and dozens of prompts, I cannot get a single request executed properly. The front-end results are highly flawed, and the back-end is incomplete and produces poor outcomes.
I would like to be able to edit a .php file and have my requested front-end and back-end updates implemented precisely where I specify, without altering any unrelated parts of the code — just as it used to work flawlessly in the past. I hope to hear from you soon that I will once again be able to use V0 as before. I look forward to your response. @jacobparis
