Technical writing.https://dperez.com/https://dperez.com/posts/who-owns-what-your-agent-learns/https://dperez.com/posts/who-owns-what-your-agent-learns/Agents that learn from their own work are already in production. Whether that learning lives in the model's weights or in the harness around it decides who can inspect, revert and keep it.Sun, 01 Nov 2026 00:00:00 GMTagentsaiharnesshttps://dperez.com/posts/php2shell/https://dperez.com/posts/php2shell/AI helped find PHP2Shell, though less than the headlines imply. The sharper lesson is what happens when bugs get cheaper to find and fixing them stays slow.Sat, 24 Oct 2026 00:00:00 GMTsecurityaiphphttps://dperez.com/posts/the-bottleneck-moved-to-the-merge/https://dperez.com/posts/the-bottleneck-moved-to-the-merge/Coding agents can produce code faster than an organisation can absorb it. Source control, review and CI each break differently, and Anthropic's Scion shows the order.Sat, 17 Oct 2026 00:00:00 GMTagentssource-controlcihttps://dperez.com/posts/build-for-the-model-you-trust-least/https://dperez.com/posts/build-for-the-model-you-trust-least/Satya Nadella argues enterprises should treat frontier models as insider risks. That moves the hard engineering into the system around the model, where a 1972 security idea still applies.Sun, 11 Oct 2026 00:00:00 GMTagentssecuritygovernancehttps://dperez.com/posts/the-agent-just-stopped-living-in-the-chat-window/https://dperez.com/posts/the-agent-just-stopped-living-in-the-chat-window/Meta put an agentic model on a single consumer GPU, while xAI and OpenAI gave their agents computers in the cloud. Neither yet gives enterprises an agent identity they can govern.Sat, 03 Oct 2026 00:00:00 GMTagentsmodelsidentity