This blog post discusses how Materialize has successfully utilized LLM-based coding agents to find bugs in existing code and open pull requests since the release of Anthropic's Opus 4.6. It explores considerations and lessons learned from their system but guides readers to the Materialize blog for further reading.