Tencent AI LAB - Self-Improving Agent

We present the code of our academic papers of self-evolving agent:

• WebEvolver: WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model, EMNLP 2025
• WebCoT: WebCoT: Enhancing Web Agent Reasoning by Reconstructing Chain-of-Thought in Reflection, Branching, and Rollback, Findings of EMNLP 2025.
• VScan: VScan: Rethinking Visual Token Reduction for Efficient Large Vision-Language Models, TMLR 2026

Friendly links to follow-up works:

• Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training.