⌘K
Samssharma 7158 Hi Amy, this is great in principle, but it’s hard to get excited about new v0 capabilities when the current v0 editor has been broken since the last upgrade. The bug has already been reported (via feedback channel and through community), yet there’s still been no meaningful update or ETA for a fix. Before promoting the next big workflow, it would be good to see the existing core experience working reliably again — and to hear clearly from the team about when that’s actually going to happen. https://community.vercel.com/t/v0-editor-issues/49664
Samssharma 7158 Any news on this ?? I don’t think anyone is even looking at this and so I’m going to escalate it on Twitter and other forums.
Sumonmaity to the team: Disabling manual code editing in the latest update is a massive step backwards. Turning the code editor into a read-only viewer and forcing users to prompt the AI for a single CSS tweak, typo fix, or one-line logic change is completely counterproductive. It wastes user credits, burns time waiting on generations, and risks the AI hallucinating or modifying unrelated code. We subscribe to the v0 Plus plan to build faster, not to lose control over our own files. AI should assist developers, not lock us out of our code. If direct inline code editing isn't restored immediately, there is zero reason for paying developers to stay when tools like Dyad, Bolt respect our workflow. Bring back the code editor
NextBlock™ CMS Just submitted our application for **NextBlock CMS**! 🚀 NextBlock is an open-source, full-stack CMS framework built natively for **Next.js 16 (App Router + RSC)** and **Supabase (PostgreSQL JSONB)**. We’ve focused on: * **Zero-Config 1-Click Deploy:** Automated Supabase database migrations on build. * **100/100 Lighthouse Defaults:** Tiptap visual editor renders PostgreSQL JSONB directly into React Server Components, eliminating client hydration overhead and HTML sanitization taxes. * **Cortex AI & Native MCP:** Model Context Protocol server support (`/api/mcp`) for deterministic, schema-constrained block synthesis. * **GitHub:** https://github.com/nextblock-cms/nextblock * **Live Sandbox:** https://cms.nextblock.dev Looking forward to the review process and collaborating with the Vercel DX team! 🧱
Axyl1410 Hi! I applied for the Summer 2026 Vercel OSS Program, but I’m not sure whether my application was successfully submitted. I also didn’t receive a confirmation email. Is there any way for applicants to verify whether their application was received, or is there normally no confirmation email after submitting the form? I’d like to avoid submitting a duplicate application if my original submission went through. Thanks!
Amy Egan If you reached the final page that says, "Thank you," then your application was received :) You'll get an email about a week after applications close letting your know whether or not your project was accepted
Ahmad Sajid This is a really fun twist on the usual snake concept. Making the player the one being chased gives it a nice sense of pressure right away. The 5-second spawn interval feels like a good starting point, but I could see the difficulty becoming much more interesting if the interval gradually decreases as the score increases. I also like the Top X% idea, especially with the global leaderboard. It gives the short runs a bit more purpose and makes you want to immediately try one more time. The React/Canvas \+ Supabase setup is a nice combination for a small browser game too. I’d definitely be interested in seeing things like different virus behaviours or temporary power-ups added later.
Physikz2 Thank you for the feedback Ahmad I really appreciate it and yes i agree with some of your points as Ive considered them myself. If the global counter does get higher maybe I will introduce new things here and there
Himu Mitra Update: recurring Jio DNS failure, with new affected-connection evidence (screenshots dated 27 September 2026; displayed local times approximately 00:46–01:11, timezone not independently verified). The affected connection is reported to be in Chennai, India. Later controlled comparisons show Airtel working; this report now focuses on Jio. On the same affected Jio hotspot / Windows laptop: 1. nslookup -type=A www.hhcare.eu.com Default resolver 10.191.189.227 returns Server failed / SERVFAIL. 2. nslookup -type=A www.hhcare.eu.com 198.51.44.13 Succeeds: 64.29.17.1 and 64.29.17.65. 3. nslookup -type=A www.hhcare.eu.com 198.51.45.13 Succeeds: 216.198.79.1 and 64.29.17.1. 4. Google Public DNS (8.8.8.8) resolves the custom hostname. Changing Android Private DNS to dns.google immediately restores access to the main domain on the same Jio phone. 5. Earlier, from the affected Jio connection: curl.exe -I --resolve www.hhcare.eu.com:443:216.198.79.1 --connect-timeout 8 --max-time 15 https://www.hhcare.eu.com/login returned HTTP/1.1 200 OK with Server: Vercel and X-Matched-Path: /login. This establishes reachability of that endpoint, not complete authentication coverage. Additional resolver comparisons: - NS eu.com via 10.191.189.227: timeout. - NS hhcare.eu.com via 10.191.189.227: Server failed. - Control NS google.com via 10.191.189.227: timeout too, so the NS-query failure is not evidence of an eu.com-specific block. - NS eu.com via 8.8.8.8: succeeds, returning ns1/ns2/ns3/ns4.centralnic.net. - nslookup -vc -type=NS eu.com 10.191.189.227: ends with Unspecified error. This generic error does not by itself prove TCP filtering. Public checks from a separate, unaffected connection found matching Vercel delegation at all four eu.com parent nameservers, successful A/SOA answers over UDP and TCP at both Vercel authoritative servers, and no child DS answer for hhcare.eu.com. These are not tests of the Jio recursive servers’ outbound path. The domain currently uses ns1.vercel-dns.com / ns2.vercel-dns.com; earlier CNAME observations in the initial post are historical, not a claim about the current wire response. Could the DNS/network team investigate authoritative reachability, response handling, filtering/rate limits or interoperability for Jio recursive-resolver traffic to this zone? 10.191.189.227 is the private resolver address visible to the subscriber, NOT the resolver’s public egress IP. We do not yet have the recursive backend/egress IP or EDE diagnostics. Please advise what operator-side evidence is needed and whether there is a supported remediation without requiring every visitor to change DNS. We are also pursuing Jio DNS support. Root cause remains unconfirmed; a universal Jio block is not established.
Himu Mitra Configuration update for investigators: at approximately 2026-09-26T19:53:21.965Z we changed the authoritative DNS delegation from Vercel to Cloudflare as a controlled diagnostic experiment. New nameservers are brian.ns.cloudflare.com and serenity.ns.cloudflare.com. Website hosting remains on the same Vercel deployment, with Cloudflare proxying OFF (DNS-only). The previous test results in post 2 were collected BEFORE this migration. We preserved the original Vercel zone for rollback, copied mail/verification/CAA records, and verified the new authoritative records on both servers. All four eu.com parent servers and Google/Cloudflare public resolvers now show the new delegation. Vercel still reports Valid Configuration, and HTTPS /login returned 200 in our check. Affected-Jio retesting without Private DNS/VPN is still pending. This is not a claim that the carrier issue is resolved. Please retain the earlier evidence when investigating the original recursive/authoritative path.
Larstraningsrapporter 8359 Same issue here. The ChatGPT ↔ Vercel MCP integration worked normally for me until yesterday evening, then suddenly stopped working across all ChatGPT chats. In my case: * list_teams returns teams: [] * all three Vercel projects on my account are affected * OAuth/reconnect succeeds * uninstalling and reinstalling/reconnecting the Vercel integration did not help * my Vercel dashboard works normally * all projects are visible and deployments continue to work * Hobby usage is far below all limits So this looks like the same OAuth/team-scope/MCP authorization problem described above, not a project-specific or quota issue. Happy to provide any additional details if useful.
Khaled0ezz 6568 I see the same `list_teams` result on a Hobby workspace: `{"teams":[]}`. The ChatGPT connection is active, and Vercel Account Settings lists “ChatGPT by Vercel MCP” with permission to read team data. `list_projects()`, `get_project({idOrName:"prj_xxx"})`, and `list_deployments({projectId:"prj_xxx"})` succeed without `teamId`; the project reports `accountId:"team_xxx"`. Supplying that `teamId` to the same calls returns 403 with “re-authenticate to this scope.” Team SAML and enforced 2FA are disabled. Reconnecting and reinstalling did not help. Is this team scope intentionally unsupported on Hobby, or can the MCP team investigate the effective OAuth grant? I can provide exact IDs privately.
Jbj11unh Additional diagnostic: I bypassed list_teams and called list_projects directly using my actual Vercel Team ID. The MCP request returned 403 Forbidden with: “Not authorized: Trying to access resource under scope jbj11unh. You must re-authenticate to this scope or use a token with access to this scope.” This happens even though the ChatGPT by Vercel MCP OAuth authorization screen explicitly displayed my team (jbj11unh's projects), and I clicked Allow Access. This appears to confirm that the team scope is not being retained or applied to the ChatGPT MCP OAuth grant. Another user in this thread appears to be seeing the same behavior when explicitly providing a Team ID.
system While a member of our team prepares to jump in, you might find your answer even faster in our community resources. - Official Documentation: https://v0.app/docs - The v0 expert walkthrough: https://community.vercel.com/t/become-a-v0-expert/5981 - Watch v0 in Action: https://community.vercel.com/live#v0
NotAVibeCoder Weird. I don't see any responses to my query. That must be a bug.
NotAVibeCoder Whoa, I still can't see any responses. What is causing this bug?
Raymond If I need to make a tiny change like: max height: max-h-[calc(100vh-7rem)] → max-h-[calc(100vh-10rem)] I can make that change myself in a few seconds. Instead, I now have to spend time explaining the change to AI, wait for it to process the request, consume credits, and then verify whether it actually made the change I wanted. By the time I finish typing the prompt, I could have already fixed the code myself. This is not making development faster. It is adding an unnecessary layer between the developer and their own source code.
TestVia Dev Just changed one value, which cost me 0.653 credits. Before this update, I was able to manually update this value. Who exactly decided to disable this editing feature? Did anyone ask our actual users?
Sumonmaity To the @v0 & @vercel team: Disabling manual code editing in the latest update is a massive step backwards. Turning the code editor into a read-only viewer and forcing users to prompt the AI for a single CSS tweak, typo fix, or one-line logic change is completely counterproductive. It wastes user credits, burns time waiting on generations, and risks the AI hallucinating or modifying unrelated code. We subscribe to the v0 Plus plan to build faster, not to lose control over our own files. AI should assist developers, not lock us out of our code. If direct inline code editing isn't restored immediately, there is zero reason for paying developers to stay when tools like Cursor, Bolt, and Lovable respect our workflow. Bring back the code editor.
Raymond I’d strongly recommend bringing back the ability to manually edit code in the Code view. Not every code change needs AI. For small, precise changes—such as adjusting a CSS value, changing a string, fixing a typo, or making a minor logic change—users should be able to edit the code directly without consuming AI credits. Requiring a conversation for every code change has several drawbacks: - Small changes unnecessarily consume credits. - AI-generated changes may modify more code than intended. - The final result is not always exactly what the user requested. - Developers lose direct control over the source code. - Simple one-line fixes become unnecessarily slow and inefficient. AI should make coding faster, not become the only way to modify code. A better workflow would be to support both: AI for generating, refactoring, and complex changes + direct code editing for small, precise changes. The ability to directly edit the source code is important to me as a developer. If manual code editing is permanently removed, I may unfortunately have to reconsider my subscription and evaluate migrating my projects to another platform that provides both AI-assisted development and direct code editing.
NotAVibeCoder Wait, what? No, this is clearly a bug. When are they going to fix it?
DWE13123 When is the v0 code editor going to be available again?? i need to do some changes manually, no with AI