Today's trail

VibeGame: Vibe Your Dream Game -- An open-source self-evolving multi-agent framework with an AI-Native game engine that turns your natural language into a fully playable 2D web game and edit it any way you want, built on Claude Code & Codex.

Harness-Delta Attribution: Dissecting Gains From Harness Evolution

Code for 'Certification of Real Images through Calibrated Content Authentication' (Hashmi et al., 2026): detector benchmark, PGD robustness, and the calibrated resynthesis / Authenticity-Index certifier.

AMLworld-Compact: Importance-Weighted Downsampling for Cost-Effective LLM Evaluation and Error Diagnosis, preprint: https://arxiv.org/abs/2610.13627