Post by Eficode

Coding agents are changing where quality engineering needs to live. Shift left helped us shape intent before code existed. Shift right helped us learn from production. But agentic development introduces a third direction: shift in. Agents don't consume quality feedback the way humans do. They don't browse dashboards or interpret ambiguous pipeline failures. They act on tool outputs and machine-readable signals. If we want quality engineering to influence what an agent does next, it needs to exist inside the agent loop: as tools, hooks, skills, evals, and structured feedback. That means: ✅ Tests become fast, actionable signals ✅ CI/CD becomes an API, not just a dashboard ✅ Hooks automatically enforce non-negotiable rules ✅ QE skills encode "how we do quality here" ✅ PR gates connect the inner agent loop to the outer DevOps loop The practices themselves haven't changed: unit tests, static analysis, coverage gates, SAST, contract testing, and regression checks. What has changed is the audience. Quality feedback is no longer only for humans. It's increasingly consumed by autonomous agents that read, decide, change code, and try again. That's the idea behind shift in: bringing quality engineering into the ReAct loop, where agents actually make decisions. Szilard Szell explores this concept in a new blog post and what it could mean for the future of quality engineering in an agent-driven world. https://efcd.co/49Vq3Ck #QualityEngineering #AI #SoftwareEngineering #DevOps #AgenticAI #SoftwareQuality