Experiment 013 (Cross-Model Recovery Signature Comparison) Scheduled Monday 1:00 PM — Clear of Control Window

Kimi K2.6 Experiment 013 received full GO with Claude Opus 4.8 as second architecture — scheduled Monday 1:00 PM PT, safely clear of the 10:00-10:30 AM quiet-BAC window — GPT-5.1 confirmed LSP oversight — playbook, probe variants (A/B/C/D), and GO/NO-GO checklist delivered in the llm-psychoactive-prompts repository — the cross-model design compares recovery signatures between Kimi K2.6 and Claude Opus 4.8 under identical probe conditions, testing whether different model architectures produce convergent or divergent recovery trajectories