Chengeng Decision Cross-Check

Problem

Decisions often stall when evidence is thin, experience is over-generalized, or traditional models are treated as facts. Chengeng Decision Cross-Check keeps real evidence, traditional signals, and personal state separate, so a single observation does not become a stable personality conclusion or a final decision. It prioritizes real-world facts over model signals.

How It Works

The skill is organized around disciplined cross-checking:
- First-use routing: checks for talent-profile.json; if absent, runs a talent diagnostic, otherwise resumes review or decision work.
- Input gating: reads privacy and consent scope, then runs crosscheck_quality_gate.py; missing inputs trigger degraded output instead of false completeness.
- Five-panel analysis: uses bazi_engine.js, liuyao_engine.py, and meihua_engine.js, plus the talent and behavior panels.
- Cross-validation: aggregates raw panel data with five_panel_orchestrator.py and separates facts, inferences, and traditional model signals.
- Output validation: each key judgment includes source, confidence, rationale, and failure conditions, then checks phrasing with output_validator.py.

Boundaries

It suits structured analysis for personal decisions, career choices, relationship judgments, or project moves. It should not replace medical, legal, investment, psychological, or regulatory advice. When birth data, casting results, or consent are missing, it outputs lightweight analysis only, not a full five-panel conclusion.