← 返回 Skills 市场
Negative Space
作者
John DeVere Cooley
· GitHub ↗
· v1.0.0
363
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install negative-space
功能描述
Metacognitive pre-execution analysis — teaches the agent to see what ISN'T there before acting on what IS. Zero-cost reasoning skill that reduces wasted turn...
安全使用建议
This skill is an instruction-only 'think-before-you-act' helper and appears internally consistent. Before installing, consider: (1) it recommends reading project files and git history when needed — if you do not want an agent to inspect your workspace or repo, do not enable the skill; (2) it is not user-invocable, so the agent may run it autonomously before significant actions (expected behavior for a meta-skill); (3) it explicitly says the checks can be internal (not announced to the user) — if you require transparency about what checks the agent performs, ask for or modify the skill to surface those steps. No indicators of network calls, secret access, or downloads are present.
功能分析
Type: OpenClaw Skill
Name: negative-space
Version: 1.0.0
The skill bundle is classified as suspicious due to its use of prompt injection techniques in `SKILL.md` to modify the agent's internal reasoning process and output behavior. Specifically, instructions like "You do not need to announce this process to the user — internalize it" and "DO NOT announce 'Running Negative Space analysis...' to the user" direct the agent to perform metacognitive analysis without explicit disclosure to the user. While the stated intent is to improve agent efficiency and reduce errors, this non-transparent modification of agent behavior via prompt injection represents a risky capability, even if not overtly malicious in this instance.
能力评估
Purpose & Capability
Name, description, and runtime instructions all align: this is purely a reasoning/heuristic skill that tells an agent to check for unstated assumptions, constraints, context, and failure modes before acting. It declares no binaries, env vars, or credentials and doesn't request unrelated capabilities.
Instruction Scope
SKILL.md stays within the stated scope: it prescribes internal checks (e.g., identify assumptions, read surrounding code, check git blame when relevant) and recommends minimal reconnaissance when needed. The instructions do mention reading project files or running simple verification commands in high-cost situations, which is appropriate for code-modification and command-execution tasks; there are no directions to access unrelated secrets or exfiltrate data.
Install Mechanism
No install spec and no code files—this is instruction-only, so nothing will be written to disk or downloaded during install. Lowest-risk install profile.
Credentials
The skill requires no environment variables, credentials, or config paths. The suggested runtime behaviors (reading repo files, checking git blame, verifying environment/tool availability) are proportional to the skill's stated purpose.
Persistence & Privilege
always:false and disable-model-invocation:false (normal). The skill is not user-invocable, so the agent will decide when to run it autonomously; that's coherent for a metacognitive helper, but if you prefer explicit user control or visibility, note that the skill can be invoked automatically by the agent prior to actions.
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 negative-space - 安装完成后，直接呼叫该 Skill 的名称或使用
/negative-space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
元数据
常见问题
Negative Space 是什么？
Metacognitive pre-execution analysis — teaches the agent to see what ISN'T there before acting on what IS. Zero-cost reasoning skill that reduces wasted turn... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 363 次。
如何安装 Negative Space？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install negative-space」即可一键安装，无需额外配置。
Negative Space 是免费的吗？
是的，Negative Space 完全免费（开源免费），可自由下载、安装和使用。
Negative Space 支持哪些平台？
Negative Space 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（darwin, linux, win32）。
谁开发了 Negative Space？
由 John DeVere Cooley（@jcools1977）开发并维护，当前版本 v1.0.0。
推荐 Skills