emotion-detector_skill

Overview

This skill detects the primary emotion in a piece of text and returns structured metadata an AI agent can use to shape its response. It delivers emotion type, intensity, valence, confidence, a recommended response strategy, and a safety flag for high-risk messages. Use it to gauge user state before composing empathetic, de-escalating, or task-oriented replies.

How this skill works

Send a short text (up to 2000 characters) and optional context; the service analyzes linguistic cues and maps the result to a normalized emotion taxonomy. It returns a unique emotion_id, the primary and optional secondary emotions, intensity (low to critical), valence, confidence score, and a concise response strategy. If intensity and emotion indicate crisis (grief/shame/fear/despair at critical), it sets a safe_t_flag to prompt immediate safety handling.

When to use it

• Before generating conversational responses to tailor tone and content.
• When triaging incoming messages for priority or escalation.
• To inform routing decisions (e.g., hand off to human support).
• To monitor sentiment trends in user feedback or chat logs.
• When compliance or safety protocols require detection of crisis signals.

Best practices

• Check confidence before trusting low-scoring predictions; combine with conversation history.
• Respect the safe_t_flag: pause automated flows and surface emergency resources or human intervention.
• Use the response_strategy field as a prompt constraint, not a full reply—adapt to context.
• Apply rate limits and privacy controls when analyzing sensitive user content.
• Combine with intent classification to avoid misinterpreting rhetorical or sarcastic text.

Example use cases

• Customer support bot detects rising anger and switches to a calm, solution-focused script.
• Mental health check-in agent identifies critical grief and stops automated triage to show crisis resources.
• Community moderation pipeline flags hostile or disgust-laden messages for human review.
• Product feedback analysis groups high-intensity negative reactions to prioritize bug fixes.
• Sales assistant senses user hope or excitement and adjusts messaging to be more proactive.