I totally get the privacy and security risks that come with AI coding agents, and especially with something shaped like claw-coder. The part that leaves a question mark for me is “closed source because we’re testing heavily”;that’s normally the argument for opening the source, not against it. More eyes, more bugs found, more reason to trust it.

Do AI agent's ability to reason about code become more powerful when given access to knowledge graph.

Cool agent man