← 返回 Skills 市场
athola

Nm Imbue Latent Space Engineering

作者 athola · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
81
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install nm-imbue-latent-space-engineering
功能描述
Shape agent behavior through instruction framing, emotional priming, and style transfer rather than information density alone
安全使用建议
This skill appears to be a pure prompt-engineering guide and is largely coherent with its stated purpose, but do the following before installing: (1) Fix or confirm the config path mismatch — the registry/SKILL.md expect night-market.modules/*.md while the bundle contains modules/*.md; if the agent loader looks for the wrong path the modules may not be loaded. (2) When using style-gene-transfer, avoid pasting real secrets, private keys, or copyrighted source as exemplars — that can accidentally propagate sensitive data. (3) If you dispatch competitive multi-agent reviews, enforce the suggested safeguards (require evidence citations, weight findings by severity) to avoid gaming or inflated reports. (4) If you need higher assurance, ask the author to correct the metadata paths and provide a short README explaining how modules are loaded by the target agent runtime. Given the path inconsistency and these workflow risks, proceed only after addressing or accepting these caveats.
功能分析
Type: OpenClaw Skill Name: nm-imbue-latent-space-engineering Version: 1.0.0 The skill bundle contains purely methodological instructions for prompt engineering, focusing on techniques like emotional framing, style transfer, and competitive review to improve agent performance. There is no executable code, no network activity, and no evidence of malicious intent or prompt injection attacks in SKILL.md or the associated modules.
能力评估
Purpose & Capability
The skill is an instruction-only prompt-engineering module and the included module files (modules/*.md) match the stated purpose. However the declared required config paths in the registry and SKILL.md metadata are night-market.modules/emotional-framing.md, etc., while the actual files are under modules/*.md. That path mismatch is an inconsistency that could break loading or cause the agent to attempt to fetch absent resources. Other declared metadata (version in SKILL.md vs registry) also differs slightly.
Instruction Scope
Instructions are limited to prompting patterns (emotional framing, style exemplar injection, competitive review) and do not instruct access to system files, environment variables, or external endpoints. Two functional risks remain: (1) style-gene-transfer encourages injecting exemplar snippets — if the agent/user sources those from private code or configuration, secrets or copyrighted material could be copied into outputs; (2) competitive framing can create incentive pressure that leads reviewers/agents to inflate findings unless the provided anti-pattern mitigations (evidence citation, weighting) are followed. The instructions themselves do not perform exfiltration, but they enable workflows that could.
Install Mechanism
No install spec and no code files to execute — instruction-only skill. This minimizes installation risk (nothing is written or executed on disk by the skill).
Credentials
The skill requests no environment variables or credentials. This is proportionate to an instruction-only prompt-engineering skill.
Persistence & Privilege
always:false and no special privileges requested. The skill does not request permanent presence or altered agent configuration beyond using its own modules.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install nm-imbue-latent-space-engineering
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /nm-imbue-latent-space-engineering 触发
  4. 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 the "latent-space-engineering" skill, ported from claude-night-market/imbue. - Enables shaping agent behavior via instruction framing, emotional priming, and style transfer rather than increasing information density. - Introduces core techniques: emotional framing, style gene transfer, and competitive review for multi-agent scenarios. - Designed for composing dispatch prompts, guiding agent behavior, and ensuring stylistic consistency in outputs. - Includes quick reference for when and how to apply each technique.
元数据
Slug nm-imbue-latent-space-engineering
版本 1.0.0
许可证 MIT-0
累计安装 1
当前安装数 1
历史版本数 1
常见问题

Nm Imbue Latent Space Engineering 是什么?

Shape agent behavior through instruction framing, emotional priming, and style transfer rather than information density alone. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 81 次。

如何安装 Nm Imbue Latent Space Engineering?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install nm-imbue-latent-space-engineering」即可一键安装,无需额外配置。

Nm Imbue Latent Space Engineering 是免费的吗?

是的,Nm Imbue Latent Space Engineering 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Nm Imbue Latent Space Engineering 支持哪些平台?

Nm Imbue Latent Space Engineering 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 Nm Imbue Latent Space Engineering?

由 athola(@athola)开发并维护,当前版本 v1.0.0。

💬 留言讨论