← 返回 Skills 市场
140
总下载
1
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install scumbag
功能描述
渣男话术认知系统 —— 从403条语录中蒸馏的情感操控模式识别与防御分析Skill。5个心智模型 + 8条决策启发式。触发词：渣男识别、话术分析、情感操控、gaslighting、PUA识别、关系预警。
安全使用建议
Plain-language considerations before installing:
- Functionality & risk: The skill appears to do what it says — pattern recognition and defensive advice for manipulative relationship talk — and it asks for no credentials or installs. That makes its operational risk low.
- Provenance: The source/homepage are unknown. Ask the publisher where the 403-line corpus came from and whether it contains any private or identifiable messages you don't want redistributed.
- Dual-use: The SKILL.md allows simulating manipulative replies (with a required warning). Even if intended for training/recognition, that feature can be misused to craft manipulative messages. If you’re concerned, avoid enabling simulation or test that the agent includes the required warning before using it.
- Privacy: The packaged references appear to be distilled from user-provided dialogues. If you or someone in your organization provided the original corpus, confirm consent and redaction of personal data.
- Practical checks: If you install it, try benign sample prompts to confirm the skill follows its own constraints (adds the simulation warning, avoids prescriptive 'be manipulative' guidance). If the skill misbehaves or you can’t verify provenance, prefer not to enable it.
功能分析
Type: OpenClaw Skill
Name: scumbag
Version: 1.0.0
The skill bundle is a specialized roleplay and analysis framework designed to help an AI agent identify and analyze emotional manipulation patterns (gaslighting and PUA) in relationship-related text. The content, including SKILL.md and the research files in the references directory, focuses entirely on psychological pattern recognition and providing defensive advice based on a dataset of 403 quotes. There are no technical indicators of malice, such as unauthorized system access, data exfiltration, or harmful prompt injection instructions.
能力评估
Purpose & Capability
The name/description (识别渣男话术、分析与防御) matches the SKILL.md content and the included reference documents. The skill requires no binaries, env vars, or config paths — consistent with an instruction-only analytic skill. The provided references are documentation and analysis derived from a 403-item corpus, which aligns with the stated distillation origin. One provenance note: source/homepage are unknown, so the origin and quality of the 403-line corpus cannot be independently verified.
Instruction Scope
The SKILL.md defines a tight agentic workflow: classify, analyze, and advise. It explicitly permits simulating '渣男' replies but mandates clear warning labels. The instructions do not ask the agent to read unrelated files, access environment variables, or send data to external endpoints. Caveat: simulation capability is dual-use — although the SKILL.md advises caution, an agent that follows the simulation path could be used to generate manipulative messages; enforcement depends on the runtime agent respecting the 'must label' constraint.
Install Mechanism
No install spec or binaries are present; this is instruction-only and writes nothing to disk on install. That minimizes installation risk.
Credentials
The skill requests no environment variables, credentials, or config paths. This is proportionate to an analysis/heuristics skill. Note: the package includes distilled content claimed to originate from user-provided private dialogs — if the original corpus contained sensitive personal data, embedding it in the skill repository would be a privacy concern, but the skill does not request access to external secrets at runtime.
Persistence & Privilege
always is false and there is no indication the skill modifies other skills or system-wide agent settings. Autonomous invocation is allowed (platform default) but not combined with extra privileges.
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 scumbag - 安装完成后，直接呼叫该 Skill 的名称或使用
/scumbag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
v1.0.0
- Initial release of 渣男话术认知系统 (zhaman-perspective).
- Distilled 403 real-world dialogues into 5 cognitive models and 8 practical decision heuristics for recognizing and defending against manipulative relationship tactics.
- Provides a step-by-step workflow for classifying questions, analyzing manipulative strategies, and offering actionable responses.
- Details typical language patterns, high-frequency phrases, and common behavioral stages in manipulative relationships.
- Includes clear boundaries and limitations, emphasizing responsible use and cultural context.
- Targets users seeking to identify, understand, and defend against emotional manipulation in relationships.
元数据
常见问题
渣男 是什么？
渣男话术认知系统 —— 从403条语录中蒸馏的情感操控模式识别与防御分析Skill。5个心智模型 + 8条决策启发式。触发词：渣男识别、话术分析、情感操控、gaslighting、PUA识别、关系预警。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 140 次。
如何安装 渣男？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install scumbag」即可一键安装，无需额外配置。
渣男 是免费的吗？
是的，渣男 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
渣男 支持哪些平台？
渣男 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 渣男？
由 dongjiang（@dongjiang1989）开发并维护，当前版本 v1.0.0。
推荐 Skills