Refine AI outputs using cosine similarity and philosophy balance scoring.
npx skills add https://github.com/lofibrainwav/AFO_Kingdom --skill soul-refine
Please help me install this Agent Skill. Skill: soul-refine Source: https://github.com/lofibrainwav/AFO_Kingdom/tree/main/skills/soul-refine Command: npx skills add https://github.com/lofibrainwav/AFO_Kingdom --skill soul-refine
This Skill refines AI outputs to match the Kingdom's aesthetic and philosophical balance using vibe alignment.
refined = soul_refine.refine(output="Draft sermon on Trinity Score", target_vibe="royal")
High-intent search queries and answers about installing and using this skill.
Vibe alignment uses cosine similarity to measure how well your output matches a target philosophy. soul-refine scores your content against dimensions like truth, goodness, beauty, and serenity, then generates actionable refinement suggestions to close gaps between current and desired alignment.
Cosine similarity quantifies the directional match between your output and a target vibe as a numerical score. soul-refine applies this to detect misalignment in content generation and code refinement workflows, enabling precise philosophy-to-output matching rather than subjective assessment.
Yes. soul-refine handles content generation, code refinement, and decision-support workflows. It applies the same cosine similarity and philosophy balance mechanics across domains, making it suitable wherever output-to-values alignment matters.
Soul-purity scoring evaluates how well your output embodies the target philosophy across multiple dimensions. soul-refine combines this scoring with philosophy balance calculations to identify which aspects of your output need adjustment and suggest concrete improvements.
soul-refine detects misalignment through cosine similarity and philosophy balance analysis, then returns specific refinement suggestions. These actionable recommendations guide you to revise content or code to better match your target aesthetic and philosophical values.
No special dependencies are required. soul-refine accepts your output as a string and a target vibe identifier, then applies cosine similarity matching and philosophy balance scoring without additional setup or input preparation steps.
randalmurphalCompute CVSS scores and prioritize vulnerabilities with weighted ranking.
vxcozyScore, diagnose, rewrite, and re-score artifacts until quality thresholds are met.
Snowflake-LabsA concise summary of your roleplay character: name,</name> personality, and/or behavior, formatted as a single string sentence. Typically不超过50字，此处省略
discountedcookieStore 384-dimensional embeddings in Postgres and query cosine similarity with pgvector.
