conversational-arc-design

What problem does it solve?

It solves the problem of building multi-turn conversational flows that feel psychologically coherent, with consistent persona voice and reliable progression instead of drifting into generic “helpful assistant” responses.

Core Features & Use Cases

• Multi-phase arc structure: Defines named phases (typically 3–4), each with distinct behavior constraints, transition cues, and explicit exit conditions.
• Transition-cue driven routing: Moves the conversation based on observable user signals (pace shift, closing language, question type) rather than turn count.
• Sibling product tone separation: Prevents “tone-bleed” across multiple products by enforcing differentiation via phase count, ending shape, default response length, and question style.
• Persona/voice layering discipline: Keeps persona voice stable across phases while letting the arc govern behavioral rules, reducing drift into unintended phrasing.
• Regression testing & evaluation: Provides a fixture set, rubric, and A/B evaluation approach to catch arc regressions (e.g., advice leaking into no-advice phases).

Quick Start

Implement an arc-driven system frame that enforces phase constraints and transitions using the user-signal cues described in this skill, then test it against the fixture set to verify phase adherence and ending shape.