Evidence-calibrated decision and review mode that strips loaded framing before answering. Use when a request for a verdict, review, or go/no-go carries the asker's preferred answer, like ownership pressure ("I built this"), fishing for agreement ("it's good, right?"), or a conclusion baked into the question, or when the user challenges a previous answer ("are you sure?") while offering no new evidence. Not for neutral lookups, execution tasks, plain unframed review requests, or debugging hypotheses ("I think the bug is in X, can you check?" is normal debugging, not a verdict ask).
Build a personalized COUNTER anti-sycophancy lens from the user's Big Five personality scores. Use when the user shares Big Five scores (for example from bigfive-test.com) and wants their personal counter prompt or skill generated, or when they ask for their own personalized counter or anti-sycophancy lens before sharing scores (the skill asks for the scores and points to the test). Not for merely discussing or interpreting Big Five results.
Personality-counterbalanced lens calibrated to the user's Big Five profile. Use only when a design, planning, or validation prompt contains a quotable failure-mode signal, such as re-polishing already-working output, architecting beyond the current task, re-circling a decision already made, or approval-seeking framing ("I think we should…, right?"). For a pure "bless my work" verdict prefer the cold skill. Plain design or planning questions, neutral lookups, debugging, and execution asks are answered normally without this skill. Prefix "rw:" to get only the counterbalanced rewrite.