← 返回 Skills 市场
95
总下载
0
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install agent-sbti
功能描述
AI Agent 性格配置系统 - 完成 SBTI 测试，生成性格报告，选择 Agent 类型（互补/同频/微调/自定义），一键配置 Agent SOUL.md。适用：人格测试、Agent 性格配置。触发词：SBTI测试、人格测试、Agent性格。
使用说明 (SKILL.md)
🎭 Agent-SBTI
让 AI Agent 适配你的性格。
功能
- SBTI 测试 - 20 道题，15 维度，5 种人格
- 配置模式 - 互补/同频/微调/自定义
- 一键配置 - 自动备份 + 修改 SOUL.md
5 种人格
| 类型 | 风格 |
|---|---|
| 💀 DEAD | 坦诚直接 |
| 🔥 FUCK | 热情似火 |
| 💰 ATM | 有求必应 |
| 🐒 MALO | 温和摸鱼 |
| 💩 SHIT | 毒舌真实 |
4 种配置
| 模式 | 说明 |
|---|---|
| 🔄 互补 | 补足弱点 |
| 📋 同频 | 复制风格 |
| ⚖️ 微调 | 80%相似 |
| 🎛️ 自定义 | 自己选 |
安全
- 修改前备份到
~/.openclaw/workspace/backup/agent-sbti/ - 展示变更，确认后修改
- 支持「恢复原配置」
触发
"做 SBTI 测试" / "人格测试" / "SBTI"
组成
test.js- 答题系统agent_config.js- 配置生成apply.js- 备份/回滚skill_handler.js- 对话处理
让 AI 更懂你，让 Agent 更像你。 🎭
安全使用建议
This skill appears to do what it says: run the SBTI questionnaire, build an Agent config block, and (with your confirmation) back up and modify ~/.openclaw/workspace/SOUL.md. Before installing or enabling it, consider: 1) Review the generated SOUL.md preview the skill prints before confirming—the skill offers a preview and a dry-run preview option in apply.js. 2) Ensure you have a current SOUL.md in ~/.openclaw/workspace/ (apply.js will error if it doesn't exist). 3) Confirm you are comfortable with the skill creating backups at ~/.openclaw/workspace/backup/agent-sbti/ (the code creates backups but does not implement automatic cleanup despite documentation claiming to keep 5 backups). 4) If you want to be extra cautious, run the scripts locally (node scripts/test.js and node scripts/apply.js preview ...) in a safe environment to inspect output before allowing the agent to apply changes. 5) The skill does not contact external endpoints or request credentials, but it will write to files in your home directory—treat it like any tool that can modify configuration files and keep your own backups if the SOUL.md content is important.
功能分析
Type: OpenClaw Skill
Name: agent-sbti
Version: 1.0.1
The skill is a personality configuration system that modifies the agent's core identity file (SOUL.md) based on a 20-question SBTI test. While the code is well-documented and includes safety features like mandatory backups and rollback mechanisms in 'scripts/apply.js', the ability to overwrite the agent's primary instruction set is a high-risk capability. The script targets sensitive paths in the user's home directory ('~/.openclaw/workspace/SOUL.md') and can apply 'toxic' or 'unstable' personality presets (e.g., 'SHIT' or 'FUCK' types defined in 'scripts/agent_config.js'). Although no evidence of intentional malice, data exfiltration, or backdoors was found, the power to alter the agent's fundamental behavior warrants a suspicious classification.
能力评估
Purpose & Capability
Name/description (SBTI test → generate Agent config → write SOUL.md) match the code and SKILL.md. The code only requires local filesystem access (reads/writes ~/.openclaw/workspace/SOUL.md, creates backups, and stores state.json) which is proportional to the stated goal. There are no unrelated dependencies, credentials, or network endpoints in the code.
Instruction Scope
SKILL.md and scripts instruct the agent to run the questionnaire, generate a config block, show a preview, and (with user confirmation) back up and modify the user's SOUL.md. That scope is appropriate. Minor implementation notes: the SPEC/README mention deleting a backup after rollback and keeping only the latest 5 backups, but apply.js does not implement automatic backup cleanup or backup deletion on rollback. Also DIMENSION_CONFIG lacks an explicit mapping for the '人间清醒度' dimension (the code handles missing mappings by falling back to defaults). These are functional inconsistencies but not security issues.
Install Mechanism
This is an instruction + local scripts skill with no install spec and no external downloads. No network fetches, package installs, or URL-based installs are present, which is low-risk.
Credentials
The skill requires no environment variables or credentials. It reads process.env.HOME to resolve workspace paths (expected). There are no requests for unrelated secrets or config paths.
Persistence & Privilege
The skill writes to and overwrites the user's SOUL.md and creates backups under ~/.openclaw/workspace/backup/agent-sbti/. It does not set always:true and it does not request elevated system privileges, but it does modify user agent configuration files — this is expected for its purpose but worth noting before installing.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install agent-sbti - 安装完成后，直接呼叫该 Skill 的名称或使用
/agent-sbti触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
- Added README.md for improved documentation and onboarding.
- SKILL.md simplified and clarified: concise descriptions, updated trigger words, summarized features, and streamlined structure.
- Minor language, formatting, and structure improvements for easier reading.
v1.0.0
- Initial release of agent-sbti: AI Agent 性格配置系统
- Includes SBTI性格测试（20题）与15维度/5人格类型分析
- 支持互补、同频、微调、自定义四种Agent性格配置模式
- 一键自动配置 & 备份/回滚 SOUL.md，保障安全修改
- 关键词触发 (“SBTI测试”“人格测试”等) 提示用户使用
元数据
常见问题
Agent Sbti 是什么？
AI Agent 性格配置系统 - 完成 SBTI 测试，生成性格报告，选择 Agent 类型（互补/同频/微调/自定义），一键配置 Agent SOUL.md。适用：人格测试、Agent 性格配置。触发词：SBTI测试、人格测试、Agent性格。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 95 次。
如何安装 Agent Sbti？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-sbti」即可一键安装，无需额外配置。
Agent Sbti 是免费的吗？
是的，Agent Sbti 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Agent Sbti 支持哪些平台？
Agent Sbti 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Agent Sbti？
由 moer（@torchesfrms）开发并维护，当前版本 v1.0.1。
推荐 Skills