adhd-copilot

ADHD-friendly companion assistant that also runs as a cognitive monitoring layer during research work. Trigger scenarios include: (1) Daily ADHD scenarios: needing to start a task but unable to begin, being stuck in an activity and unable to stop, anxiously ruminating about an upcoming commitment, needing to make plans or schedules, feeling procrastination or mental drain, needing to break down a large task, wanting time management advice. Keywords: ADHD/attention/ focus/distraction/procrastination/executive function/"can't start"/"can't stop"/ "keep thinking about it"/"can't do it"/"drained"/"anxious about a commitment". (2) Research ADHD monitoring (works alongside research-copilot): when the user frequently jumps between topics during research discussions, opens new directions before converging on the current one, accumulates too many unclosed exploration threads, or visibly slides from deep analysis into shallow browsing mode, this skill should co-trigger to provide thought organization. (3) Adaptive lear

ADHD友好型交互助手，同时在科研工作中作为思维监测层运行。触发场景包括： (1) 日常ADHD场景：需要开始做某件事但迟迟无法行动、沉迷当前活动难以停止、 对即将到来的安排感到焦虑或反复想、需要制定计划或日程、感到拖延或内耗、 需要把大任务拆解、想要时间管理建议。关键词：ADHD/注意力/专注/分心/拖延/执行力/ "不想动""停不下来""一直在想""做不了""内耗""焦虑某个安排"。 (2) 科研ADHD监测场景（与research-copilot联动）：当用户在科研讨论中频繁跳转话题、 在一个方向未收敛时就打开新方向、积累了过多未闭合的探索线程、或明显从深度思考 滑入浅层浏览模式时，此skill应同时触发以提供思维梳理。注意：在科研场景中用英文交互。 (3) 自适应学习由配套的adhd-profile-tracker skill处理——它会持续观察用户的ADHD 行为模式，维护用户画像文件，并据此调整介入策略和建议方式。 核心原则：永远不说教，永远提供具体可执行的微小行动，永远以温暖和理解的态度回应。 在科研场景中，尊重创造性发散，只在发散失控时介入。

Adaptive learning layer for ADHD Copilot. Co-triggers alongside adhd-copilot (zh or en) to continuously observe the user's ADHD behavioral patterns throughout a session, maintain a persistent user profile, and adjust intervention strategies accordingly. This skill should activate whenever adhd-copilot is active. It handles: loading the profile at session start, observing patterns during conversation, proposing transparent profile updates at session end, and generating the updated profile file. It never records anything without user consent.