AI code agentssoftware engineeringpublished Aug 20, 2026 · arXiv:2608.18389
A Jagged Frontier: Evaluating Robustness of Code Agents to Semantics-Preserving Transformations
coding agentsreinforcement learningpublished Aug 19, 2026 · arXiv:2608.17393
LEGO-RL: Harness-Native Reinforcement Learning for Coding Agents
coding agentsevaluation methodologypublished Jul 28, 2026 · arXiv:2607.22585
The Scaffold Effect in Coding Agents: Harness Choice as a Hidden Variable in Coding-Agent Evaluation
computer-use agentsdesktop automationpublished Jul 17, 2026 · arXiv:2607.14443
Tactile: Giving Computer-Using Agents Hands and Feet