Reliable Agentic Development on a €40 Budget: Dependency-Aware Orchestration for Claude & Codex

Reliable Agentic Development on a €40 Budget: Dependency-Aware Orchestration for Claude, Codex, and Human-in-the-Loop Most agentic coding s show the happy path: AI gets task, AI writes code, done. What they don’t show is who decides what the tasks are. Or what happens when a task is marked Done but the file never got created. Or when the agent silently hangs on an auth error. Or when a dependency chain falls apart and you don’t find out until three tasks downstream have already run on bad assumptions. I built a Python orchestrator that handles the full lifecycle. Codex does the building.