github:EleutherAI/rh-indicators · first seen 2026-10-31 · source · badge
Reproduces the full prefill sensitivity analysis pipeline for reward hacking indicators. Use when evaluating how susceptible model checkpoints are to exploit-eliciting prefills, computing token-based trajectories, or comparing logprob vs token-count as predictors of exploitability.
Listed, not yet scored. This skill has no public model test yet. It is in the scoring queue; submitted skills go first, then the most-installed.