What problem does it solve? AI agents often respond in ways that miss a user's actual working style: they skip offering choices before acting, over-explain, act without approval on sensitive changes, or repeat the same response mistakes. This Skill maintains a response contract based on the user's explicit feedback and repeated work patterns, so answers, reviews, plans, and completion reports match how the user actually works. ## Core Features & Use Cases - Three-option decision pattern: Presents three concrete options with a recommendation whenever next actions, approval units, scope, or verification boundaries are at stake, including slash-based parallel selection like 1/2. - Feedback loop with local logging: Records response misses and out-of-options answers as local feedback files with evidence grades (explicit, repeated, tentative, rejected), and only promotes stable patterns to long-term rules after explicit user approval. - Task silo execution defaults: Treats task execution requests as silo preparation, handling status-update PRs, goal.md creation, repo cloning, and branch setup in one flow. - Use Case: A user says "you should have given me options." The Skill acknowledges the miss, logs a feedback file under local/personality-feedback-log/feedback/active/, adjusts the response rule for the current thread, and later bundles multiple feedback entries into a personality update report for user-approved promotion. ## Quick Start Ask the agent to adjust its response style to your feedback, for example: "From now on, give me three options with a recommendation before making any source changes, and log this preference."
