← 返回 Skills 市场
Multi-Angle-Thinking
作者
hasanen groof
· GitHub ↗
· v1.0.1
· MIT-0
197
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install thinking-engine
功能描述
A deep thinking engine that first gathers real global data via web search, then analyzes through 11 lenses (philosophical, ethical, practical, historical, ps...
安全使用建议
This skill appears internally coherent and matches its description. Before installing: (1) If you enable the OpenClaw hook, know it will scan incoming prompts for trigger phrases and inject a reminder — this can cause the agent to pursue deep analysis when prompts contain those keywords, so review/disable the hook if you want stricter manual control. (2) The skill uses web_search/web_fetch — ensure your platform's search/fetch tools respect paywalls and privacy (the skill explicitly says not to fetch private/paywalled data). (3) Be cautious when simulating real named people: the SKILL.md requires labeling such simulations as fictional inference, but you should avoid asking it to fabricate private-person behavior. (4) No credentials or downloads are requested by the skill; treat it like an instruction-only tool with optional local hooks and test it with non-sensitive topics first.
功能分析
Type: OpenClaw Skill
Name: thinking-engine
Version: 1.0.1
The 'thinking-engine' skill bundle is a sophisticated prompt-engineering framework designed for deep analytical simulation. It uses standard OpenClaw extension points, including a JavaScript hook (handler.js) and a shell script (activator.sh), to inject methodological instructions into the agent's context when triggered by specific keywords. The code and markdown instructions are transparent, align strictly with the stated purpose of multi-angle analysis, and include explicit guardrails to limit web searches and protect privacy.
能力评估
Purpose & Capability
The name/description (multi-angle analysis + simulation) match the SKILL.md, README, and reference guides. Required resources are minimal (no env vars, no external binaries) and the instructions explicitly call for web_search/web_fetch which is coherent with the declared purpose.
Instruction Scope
SKILL.md prescribes a bounded pipeline (max 6 web searches, avoid paywalled/private data, label real-person simulations). That scope aligns with the stated purpose. One point to watch: the included hook (hooks/openclaw/handler.js) automatically scans prompts for trigger phrases and injects a reminder; while the SKILL.md says 'Only invoke on explicit user request,' the hook's trigger is keyword-based and may cause reminders (and thus prompt the agent toward invocation) on any prompt containing those words, causing potential false-positive activation. The skill text does not instruct collecting sensitive private data.
Install Mechanism
There is no remote download/install spec; the skill is instruction-first. The only code files are a small hook handler and an activator script; both are short, readable, and do not fetch external code or create network endpoints. No extract-from-URL or third-party installer is used.
Credentials
The skill requests no environment variables, no credentials, and no config paths. Its runtime behavior (web_search/web_fetch) relies on platform-provided web tools; nothing in the package asks for unrelated secrets or system access.
Persistence & Privilege
always:false and no privileged flags are set. However, the package includes optional hooks (handler.js and activator.sh) that, if installed/enabled by the user, will add prompt-monitoring behavior to the agent (injecting reminders on matches). This is a limited, intentional presence and not an 'always' forced installation, but users should be aware enabling hooks changes the agent's prompt-processing behavior.
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 thinking-engine - 安装完成后，直接呼叫该 Skill 的名称或使用
/thinking-engine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
**thinking-engine 1.0.1 — first minor update introducing guardrails, structure, and new docs**
- Added detailed usage guide, simulation guide, lens reference, practical examples, and usage hooks.
- Introduced explicit guardrails: web search/query limits, user consent requirements, privacy and sensitivity rules.
- Now only triggers pipeline on explicit user request (no self-invocation on simple queries).
- Clarified simulation boundaries for real individuals; added stronger privacy and safety statements.
- Included new supporting files: README, usage examples, activation scripts, and documentation for extensions.
v1.0.0
thinking-engine 1.0.0 — Initial Release
- Introduces a three-layer deep thinking engine: intelligence gathering, multi-lens analysis, and realistic simulation.
- Analyzes complex questions using 11 analytical lenses, from philosophical to counterfactual angles.
- Always grounds analysis in real-world data sourced via web search; sources are shown.
- Builds realistic narrative simulations with data-driven characters and surprising outcomes.
- Response structure adapts language and detail based on the input and cultural context.
- Designed to surface hidden biases, challenge assumptions, and offer actionable insights for both Arabic and English users.
元数据
常见问题
Multi-Angle-Thinking 是什么？
A deep thinking engine that first gathers real global data via web search, then analyzes through 11 lenses (philosophical, ethical, practical, historical, ps... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 197 次。
如何安装 Multi-Angle-Thinking？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install thinking-engine」即可一键安装，无需额外配置。
Multi-Angle-Thinking 是免费的吗？
是的，Multi-Angle-Thinking 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Multi-Angle-Thinking 支持哪些平台？
Multi-Angle-Thinking 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 Multi-Angle-Thinking？
由 hasanen groof（@h7g7）开发并维护，当前版本 v1.0.1。
推荐 Skills