Smart Follow-ups

**Judgment:** A focused, well-scoped utility skill that meaningfully improves chat UX by offering contextual follow-up questions. It’s especially useful in educational, exploratory, and support contexts where users may not know what to ask next. **What it does well** - Adds a consistent `/followups` (plus aliases/natural language) mechanism to surface three follow-up suggestions: **Quick**, **Deep Dive**, and **Related**. - Uses the existing conversation context and current model/auth, so it feels native to the chat and requires minimal setup. - Channel-aware interaction (buttons vs numeric replies) makes it broadly deployable across Telegram/Discord/Slack and SMS-like channels. **Key risks / limitations** - Safety is entirely inherited from the underlying model and policy; suggested questions may sometimes steer into sensitive or unwanted topics if the base model would do so. - No explicit controls mentioned for customizing or constraining the types of follow-ups (e.g., domain restrictions, compliance filters). - Reliability details (error handling, behavior on long/complex conversation histories, rate limits) aren’t documented, so operational robustness isn’t guaranteed. **Recommended scenarios** - **Learning & onboarding:** Tutoring, technical education, or product training chats where structured exploration is valuable. - **Developer & power-user chats:** Technical support, DevOps, and programming assistants where “Deep Dive” and “Related” questions help uncover relevant details. - **General-purpose assistants:** Any assistant where you want lightweight, low-friction guidance for “what to ask next” without redesigning the core bot logic.