Skill UI

This技能 is used when analyzing complex propositions, debating, or reviewing arguments within the CrossFrame Suite. It helps users deconstruct claims into testable arguments, identifying hidden premises, strongest counterarguments, evidence requirements, and conditions for withdrawal, ensuring conclusions are robust and non-judgmental.

This skill is designed for generating structured, insightful, and boundary-aware short answers, such as reader replies, editor responses, or consultation advice. It structures analysis by defining fact boundaries, providing structural judgment, offering necessary critique, and suggesting safe, low-risk next steps. It is primarily activated by the CrossFrame Suite, ensuring the output maintains a canonical Chinese semantic layer while providing professional, actionable advice.

A relationship-aware gift engine that determines whether a personalized gift should be created, then uses a five-stage creative pipeline (editorial judgment, synthesis, concept generation, visual strategy, rendering) to produce H5, image, or video gifts with anti-repetition and taste profiling.

Measure rendered UI layout with reproducible spatial metrics and a separately assessed usability review. Distinguishes objective measurements from vision judgments and user testing for AI-generated interfaces.

Learn to design Jev-assisted workflows using collected use cases, references, and examples. Adapt patterns for classification, scoring, or candidate selection, including batch judgments and agent checkpoints.

Choose the next legal action in browser, desktop, game, or simulation. Provides structured decision-making for agents, with options for real AI judgment or simulation. Includes safety checks, mode selection, and workflow examples.

This skill evaluates supplied outputs against explicit criteria, including code-change reviews and authorized safety evaluations. It supports evidence-backed review leads, rubric judgments, and batch transcript evaluation for existing outputs or observed behavior. It helps users judge results without merging or running targets. Suitable for design requests, code reviews, and safety workflows. Users can choose real API calls or simulation modes for independent outcome validation.

Classify and prioritize user-defined inboxes, support tickets, feedback, or records using custom criteria. Supports bulk parallel judgments with smoke_test pilot, review queues, and labels without automatic replies or mailbox changes.

Routes enumerable judgment steps (did it work, which option, how risky) to the Jev judgment model via MCP tools, batching questions into single calls and handling escalated verdicts. For agents needing fast typed decisions without token generation.

Guides an AI agent to give non-judgmental, evidence-informed quit-smoking support: quit planning, craving check-ins, slip debriefs, optional consent-based logging, and escalation to clinicians or quitlines. Not medical care.