咨询级智能分析系统

咨询级智能分析系统。将碎片化信息通过结构化思考转化为可执行洞察。 使用场景：(1) 深度行业/市场/竞品分析 (2) 战略规划与决策支持 (3) 复杂问题根因诊断 (4) 创新方案设计 (5) 用户说"分析一下"、"深度研究"、"帮我调研"、"做个报告"时触发。 包含12大分析框架（诊断/战略/创新/决策/系统/...

Security Scan

The skill is an instruction-only 'consulting/analysis' assistant that only prescribes web searches and structured analysis using included framework templates; its requested footprint (no env vars, no installs, no binaries) matches its stated purpose.

Purpose & Capability

Name/description (consulting-grade analysis) align with the content: the instructions describe structured question understanding, web-based intelligence gathering, framework selection, analysis, validation, and report output. There are no unexpected credentials, binaries, or install steps required.

Instruction Scope

SKILL.md limits runtime behavior to asking clarifying questions, performing web_search/web_fetch style searches, applying included framework templates (references/*.md), synthesizing and citing sources, and producing reports. It does not instruct reading local files, environment variables, or sending data to unknown endpoints. The guidance to perform multiple searches and cite URLs is expected for this purpose.

Install Mechanism

No install spec and no code files beyond plain MD references. Because this is instruction-only, nothing is downloaded or written to disk by the skill itself.

Credentials

The skill declares no environment variables, credentials, or config paths. Its behavior (web searches and using the provided templates) does not require additional secrets or privileged access.

Persistence & Privilege

always is false (good). disable-model-invocation is false which allows autonomous invocation — this is the platform default and not itself a problem, but note that the skill can be invoked autonomously by the agent when triggers match (e.g., user asks for deep research).

Assessment

This skill is instruction-only and internally consistent with its stated goal of producing structured research reports. Before installing or enabling it, consider: (1) the skill will perform web searches and include URLs in reports — avoid sending confidential or proprietary text to it if you don't want that content to be used in web queries or stored in agent logs; (2) verify the agent's web_search/web_fetch implementation and privacy policy (which service executes the searches and how logs are handled); (3) because autonomous invocation is allowed by default, disable autonomous invocation or require user confirmation if you prefer to prevent the agent from starting long-running research tasks without explicit approval; (4) review generated reports and cited sources for accuracy and bias before acting on recommendations. No code or credentials are requested, and nothing in the files indicates hidden endpoints or exfiltration.

Super Analyst — 咨询级智能分析系统

核心流程（7步）

1. 问题理解 → 2. 复杂度评估 → 3. 情报搜集 → 4. 框架选择 → 5. 结构化分析 → 6. 交叉验证 → 7. 输出报告

Step 1: 问题理解

明确分析目标、范围、受众。追问模糊点，不假设。

Step 2: 复杂度评估

级别	标准	策略
L1 简单	概念解释、单维度、无需外部信息	直接答，1个框架
L2 中等	需部分外部数据、2-3维度	2-4次搜索 + 1-2个框架
L3 复杂	深度研究、多维度、战略决策	5-10次搜索 + 2-4个框架组合

执行中发现复杂度升级，自动调整。

Step 3: 情报搜集

• 中文优先：中国市场/公司/政策
• 英文优先：国际公司/学术/技术创新
• 双语并用：跨境/对比分析
• 搜索模式：web_search（快速概览）/ web_fetch（深度内容）
• 质量优于数量，动态调整搜索方向

Step 4: 框架选择

读取 references/frameworks.md 了解所有框架，根据问题类型选择：

问题类型	推荐框架
"为什么总是出问题？"	5 Whys + Pareto
"如何颠覆这个行业？"	First Principles + Porter's
"应该进入这个市场吗？"	SWOT + Porter's + Scenario
"怎么优化团队效率？"	MECE + Pareto
"未来会怎样？"	Scenario + Systems Thinking

多框架组合见 references/combinations.md。 最多选 3-4 个框架，避免信息过载。

Step 5: 结构化分析

对每个选定框架，读取 references/frameworks.md 中对应模板，严格遵循6步流程，融合搜集到的情报数据，每个框架产出 800-1500 字。

Step 6: 交叉验证

• 模式连接：不同框架的结论是否互相印证
• 矛盾化解：冲突点分析，给出判断
• 深层洞察：跨框架得出的隐藏规律

Step 7: 输出报告

统一格式：

# 分析报告：[主题]

## 执行摘要
[1-2段：直接答案 + 核心结论]

## 详细分析
[按框架分别展开]

## 跨框架整合
[模式连接 + 矛盾化解 + 深层洞察]

## 行动建议
**立即（0-3个月）：** ...
**短期（3-6个月）：** ...
**中期（6-18个月）：** ...

## 信息来源
[来源及URL]

---
Level [X] | 来源数：[X] | 框架数：[X]