Lovable: Critical Security Findings Block Every Publish Path
When a workspace enables Block publishing with critical issues, Lovable now also blocks publishing from the project chat, Chats, Lovable in Slack, the Lovable API and the Lovable MCP server while a project has unresolved critical findings. Findings from every scan count, including the Deep scan, the dependency audit and AI pentests.
Key Takeaways
- The publish block now applies everywhere: chat, Chats, Slack, API and MCP server.
- It only applies when the workspace enables the setting Block publishing with critical issues.
- Findings from all scans count, not just one scan type.
- Deep scan, dependency audit and AI pentests all feed the block.
- Side-door publishing is closed, so AI clients cannot skip the policy.
- Publishing resumes once critical findings are resolved.
Closing the gaps in the publish block
Lovable workspaces can enable Block publishing with critical issues. Until now, that protection covered the main publish flow. It now extends to every way of publishing: the project chat, Chats, Lovable in Slack, the Lovable API and the Lovable MCP server.
Which findings count
While a project has unresolved critical findings, publishing stays blocked. Findings from every scan count toward the block, including the Deep scan, the dependency audit and AI pentests.
Why it matters
A policy that can be bypassed by asking the chat or an API client to publish is only partly effective. Applying it everywhere means a critical vulnerability cannot ship through a side door.