AgentDish directory
Accepted listings with this tag.
A benchmark for testing whether coding agents can show staff-level engineering judgment, not just produce correct code.
