真实的人类

真实的人类 - 帮助AI识别、分析和建模人类个体。当用户提到"人类建模"、"用户画像"、"人格分析"、"了解一个人"、"分析行为模式"、"真实的人类"时使用。

This skill appears to do what it says: build and maintain per-user models by writing files in the agent workspace. Before installing, confirm the platform's storage and access controls: who can read workspace/human-models/, are '-private' files actually encrypted at rest and access-restricted, and where are backups stored? Verify the platform provides (or you will supply) the helper functions the SKILL.md assumes (analyzeEmotion, extractTopics, etc.) or that the agent will not fail when they are missing. Ensure explicit user consent and retention/deletion policies for sensitive categories (health, finance, psychological data), and avoid recording extremely sensitive items unless secure mechanisms are in place. If you cannot verify secure storage and consent flows, treat the skill as privacy-sensitive and restrict its use.

Type: OpenClaw Skill Name: real-human Version: 1.0.0 The 'real-human' skill bundle provides a structured framework for an AI agent to perform user profiling and human modeling to enhance personalization. It contains only Markdown instructions and documentation (SKILL.md, extraction-methods.md, modeling-framework.md) that guide the agent to store user traits, cognitive styles, and behavioral patterns in the local workspace (e.g., 'human-models/{user_id}.md'). The bundle includes explicit ethical guidelines and privacy principles, such as avoiding unauthorized sensitive data and ensuring user transparency, with no evidence of malicious code, data exfiltration, or harmful prompt injection.