relationship-reading
Analyze romantic, family, friendship, or workplace relationship dynamics from user-provided profiles, history, and optionally verified charts. Use for 合盘, 复合, 暧昧, 婚恋, 相处模式, 亲友关系, or 职场关系; not for surveillance or certainty about another person's private thoughts.
Leon-Drq · 站主发布
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add https://github.com/Leon-Drq/relationship-reading-skill/tree/d60463bc693a262522a8e000a4d8760e53021760 --skill "relationship-reading"技能说明
name: relationship-reading description: Analyze romantic, family, friendship, or workplace relationship dynamics from user-provided profiles, history, and optionally verified charts. Use for 合盘, 复合, 暧昧, 婚恋, 相处模式, 亲友关系, or 职场关系; not for surveillance or certainty about another person's private thoughts.
Relationship Reading
The useful outcome is a safer interaction experiment, not a verdict about who is destined for whom.
Input gate
Collect relationship type, participants, current status, the event or repeated pattern being examined, desired outcome, non-negotiable boundaries, and only the birth/chart data the user is authorized to share. For a 6yao.ai analysis, preserve scenarioId, profiles, and language as structured fields.
Workflow
- Separate reported behavior from interpretations of motive.
- If charts are used, calculate each independently before comparing. Do not merge fields from different systems.
- Build the analysis around attraction/resources, communication, conflict cycle, power and boundaries, repair capacity, and external constraints.
- For every chart-based claim, name the exact matching or conflicting fields. For every behavioral claim, quote or paraphrase the supplied event.
- Offer a low-risk communication test with a clear stop condition. Abuse, coercion, stalking, threats, or financial control require safety-first guidance, not reconciliation advice.
Output
Lead with the central interaction pattern. Then provide evidence for strengths, friction, unknowns, a two-sided communication script, boundary or safety note, and a 7- or 30-day review experiment. Do not assign diagnoses or claim that a third party definitely loves, cheats, returns, or changes.
Read references/evidence-contract.md and references/safety-and-privacy.md.