AURA

**Quick judgment** AURA is a focused, well-specified personality-configuration layer that’s especially valuable when you want consistent, explicit control over an agent’s tone, directness, autonomy, and sycophancy across a workspace. **What it does well** - Uses a clear HEXACO-based trait model plus style/boundary sliders to turn vague preferences into concrete, persisted settings. - Provides straightforward commands (`/aura`, `/aura show`, `/aura reset`) and generates a readable AURA.yaml that can be applied automatically at session start. - Encourages safer epistemic behavior via `always_correct_errors` and `flag_uncertainty`, and limits flattery via `max_adulation`. **Key risks / limitations** - Over-aggressive settings (high assertiveness, very low agreeableness) can produce responses that feel confrontational or overly blunt if not tuned carefully. - Misconfigured or manually edited AURA.yaml may lead to confusing or inconsistent behavior if downstream logic doesn’t validate the file. - It only governs style and interaction patterns; it does not itself add domain capabilities or task-specific tools. **Recommended scenarios** - Long-term projects where you want a stable, named “persona” (research partner, code reviewer, product manager, tutor). - Users explicitly asking to "configure personality," reduce flattery or sycophancy, or tune communication (direct vs diplomatic, independent vs permission-seeking). - Teams that want shared, documented norms for how the AI should communicate and behave within a given repository or workspace.