Model, test, monitor. A price-monitoring agent scrapes a competitor's page every night, exactly as it was built to, and one line of fine print on that page walks your cost basis straight back to them. No exploit, just an agent reading content it was told to read. We close the loop live, from a failing security test to an exported guardrail.
A new open-source library for running coding agents as remote background jobs. Define an agent declaratively, run it locally or in the cloud, and swap harnesses (Claude Code, Codex) and LLM providers without vendor lock-in, live on Zyte's own spider-writing agents.
Recording, slides and both repos go out this week.
See you shortly for meetup #3, date coming soon.