Agents
Controlled agentic CAD comparison: six unattended runs, 16 failures, 9 of which the tool never reported
ModelRift published a controlled comparison on 2026-09-12 of CadQuery 2.8.0 against OpenSCAD 2026.06.12 for unattended agent-driven part generation, running six agents over three tasks with Claude Opus 5 through Claude Code, one isolated subagent per cell, and every output STL checked by an independent parser. All six parts were printable, so the result turned on the path there: 16 distinct failures across six runs, of which 9 were never surfaced by the tool at all. The silent-failure count is the transferable number, since it measures how much an agent can be trusted to notice its own errors in a toolchain without a verification loop.
Source
↳ Follow the thread