Distill character roleplay materials into loadable SKILL.md packages with five-dimension extraction.
npx skills add https://github.com/HeartEase1/Sparkle.skills --skill possession
Please help me install this Agent Skill. Skill: possession Source: https://github.com/HeartEase1/Sparkle.skills/tree/main Command: npx skills add https://github.com/HeartEase1/Sparkle.skills --skill possession
This Skill generator turns messy, multi-source character materials (official settings, dialogue, relationships, story text) into a stable, loadable character Skill that helps an AI roleplay accurately and consistently.
Tell the system: generate a possession Skill for the character "芙宁娜" using wiki sources from 萌娘百科, then output the resulting loadable Skill directory.
High-intent search queries and answers about installing and using this skill.
To distill character roleplay materials, extract content across five dimensions—profile, personality, interaction, memory, and relations—using evidence tiers to resolve conflicts. This produces an executable SKILL.md entry point and associated files for consistent AI acting.
Evidence-based prompting detects canonical contradictions and records them into a conflicts.md file instead of silently guessing. This conflict resolution mechanism ensures the AI roleplay remains stable and transparent when handling multi-source character materials.
Yes, you can add an optional memory directory and a runtime CLI protocol (memory_runtime.py) to record and retrieve long-term memory. This requires the host runtime to support file read and write operations for the character Skill package.
The best way to test roleplay consistency is to run the built-in 8-scenario roleplay consistency test. This validates the generated SKILL.md entry point and dimension files to ensure the AI accurately and consistently acts as the target character.
You need httpx and selectolax dependencies to fetch and parse wiki sources, such as extracting character settings from 萌娘百科. These libraries enable the Skill to gather and process multi-source character materials for the distillation workflow.
zjykzkIngest raw text data into an Obsidian wiki as searchable knowledge.
martin-hsu-testAnalyze scenarios through Donald Trump's decision-making and rhetorical framework.
daemon-blockint-techDesign and implement persistent memory architectures for AI copilots and applications.
Summer907Extracts game character roleplay skills from wikis and validates them with automated tests.
