https://hlynai.com/blog/ Notes from the hlyn team on runtime containment, agent isolation, and what happens after a prompt injection gets through. en https://hlynai.com/blog/01/ https://hlynai.com/blog/01/ Sat, 12 Sep 2026 00:00:00 GMT We opened a mock clinical AI agent to three days of red-teaming. Here is what held, what didn't, and what we're not claiming.