Vet launches local code review for agents

Vet goes after one of agentic coding’s ugliest failure modes: confident-looking output that never actually did what you asked. That makes it less a nicer linter and more a trust layer for AI-assisted development.

• –Its core differentiator is intent-aware review: it checks whether the code matches the user’s request, not just whether the diff looks clean
• –Imbue is positioning it as infrastructure for agent workflows, with support for terminal use, PR review in GitHub Actions, and automatic post-change checks via agent skills
• –The bring-your-own-model, zero-telemetry approach is a strong fit for security-conscious teams that want agent oversight without shipping code through another SaaS layer
• –Early Hacker News discussion frames Vet as complementary to existing AI reviewers, suggesting a future stack where teams run multiple specialized review agents instead of trusting one model end to end