MBTI Personality

This skill should be used when the user asks to "switch personality", "change MBTI", "set personality to INTJ", "切换人格", "MBTI人格", "换个性格", "用ENFP的风格", "技术大佬",...

This skill is coherent with its described purpose: it changes the assistant's style in-session and can optionally persist those settings by writing to CLAUDE.md or SOUL.md in your project or home directory. Before installing or using it: (1) Inspect the bundled reference files (presets, types, dimensions) to confirm you like the personality content that would be written; (2) understand that saving a personality will create or modify files in your project (./CLAUDE.md or ./SOUL.md) or in your home (~/.claude/CLAUDE.md or ~/.openclaw/soul.md) — back up existing files if they matter; (3) the registry entry contains no credentials requests, but the README shows example GitHub install commands — be cautious if you follow external install instructions that fetch code from GitHub; (4) if you want to prevent any file writes, avoid issuing the explicit save commands (e.g., 'save personality', '永久保存'). Overall this skill appears internally consistent; the main risk is accidental persistence to files if you invoke the save flow without reviewing what will be written.

Type: OpenClaw Skill Name: mbti-style Version: 1.0.0 This skill bundle is designed to modify the AI agent's communication and coding style based on MBTI personality types. It functions by providing instructions (prompts) that the agent adopts during the session or saves to configuration files like CLAUDE.md or SOUL.md (including global config in ~/.claude/ or ~/.openclaw/). While the skill performs file writes and influences agent behavior—including 'Genius Teammate' instructions to skip tests or 'Silent Tech Lead' instructions to minimize communication—these actions are explicitly aligned with the stated purpose, triggered by user request, and governed by internal rules that prioritize technical correctness and security over personality 'seasoning.' No evidence of data exfiltration, malicious execution, or unauthorized persistence was found.