di-perplexity-style

v1.6.0

Perplexity 风格的搜索回答规范 + Brave Search。拆→对→连三步法,时代锚定,多情景因果,个人行动闭环。

0· 159·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for wudi488/di-perplexity-style.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "di-perplexity-style" (wudi488/di-perplexity-style) from ClawHub.
Skill page: https://clawhub.ai/wudi488/di-perplexity-style
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install di-perplexity-style

ClawHub CLI

Package manager switcher

npx clawhub@latest install di-perplexity-style
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
The name/description promise (Perplexity-style search answers using Brave Search) matches the SKILL.md instructions. The guidance focuses on search, citation, time‑anchoring and analysis steps — behaviors consistent with the declared purpose.
Instruction Scope
Instructions stay within answering/search analysis scope (three-step method, seven-step full analysis, require citing sources and freshness). Two minor notes: the SKILL.md references local files (references/market_ratios.md, references/example_analysis.md) and an external integration (agent-reach) that are not provided in the skill bundle — the agent may try to access these or fall back to other sources. Otherwise there are no directives to read system files, env vars, or exfiltrate data.
Install Mechanism
No install spec and no code files — this is instruction-only, so nothing is written to disk or executed beyond the agent's normal runtime tools.
Credentials
The skill declares no required environment variables, credentials, or config paths. It does not request secrets or unrelated service access; requested capabilities (web_search) are appropriate for the purpose.
Persistence & Privilege
always is false and the skill does not request persistent or elevated platform privileges. Autonomous invocation is allowed by default but is not combined with other concerning permissions.
Assessment
This skill appears coherent and low-risk: it is a guidance/template for producing sourced search analyses and asks for nothing sensitive. Before installing, consider: (1) Confirm your agent has a web_search/Brave Search tool available — otherwise the skill's instructions that call web_search will fail or cause the agent to substitute other data sources. (2) The SKILL.md references local reference files (references/market_ratios.md and references/example_analysis.md) and an integration (agent-reach) that are not bundled; decide whether those resources exist in your agent environment or whether the agent will fetch equivalents from the web. (3) Review outputs for accuracy and source quality — the skill instructs the agent to surface links and freshness, but the agent may still hallucinate assertions if source quality is poor. (4) If you enable autonomous invocation for agents that have access to external systems, be aware the agent may run searches and post-process results without prompting; this is normal but worth auditing in high-security environments. If any of the missing references or integrations would require credentials or external services, verify those bindings before use.

Like a lobster shell, security has layers — review code before you run it.

latestvk975wkwjsqx22t7e21jgs9tm8d85fjtq
159downloads
0stars
5versions
Updated 3d ago
v1.6.0
MIT-0

Perplexity 风格回答规范

⚡ 模式选择

模式触发词适用场景执行步骤
快速"快速分析""简要分析"简单问答、快速查证数据校验 → 结论(跳过时代锚定)
完整默认 / "深入分析"复杂事件、投资决策、地缘政治全七步

核心三步法:拆 → 对 → 连

拆:分解问题

  • 时间维度(短期/中期/长期)
  • 主体维度(地缘/能源/货币/个人)
  • 事实层 vs 解读层 vs 立场层

对:数据校验

三个问题:

1. 这个数字是绝对值还是相对值?
2. 数量级对不对?
3. 和历史均值比,是高位还是低位?

关键比率(如金油比)必须量化。异常极端数字重点关注。

连:因果串联

横向连(事件间互相影响)+ 纵向连(短期/中期/长期驱动力)


搜索规范

  1. 使用 Brave Search(web_search)
  2. 标注时效性(freshness: day/week/month/year)
  3. 区分"已确认"vs"待证实"
  4. 每个关键信息标注来源链接

完整分析框架(七步)

一、时代背景锚定

这个事件发生在什么时代背景下?这个时代的主要矛盾是什么?

必须回答:大背景 → 主要矛盾 → 时代对事件的影响

二、AI/技术核心变量

AI/技术如何影响这个事件?事件对AI发展有什么意义?

三、为什么是现在?

为什么这个时间点发生?有什么特殊性?

四、跨事件联动(多情景因果链)

每条链条显性化:"如果X则A,如果Y则B" 区分"安全溢价段"和"收紧反应段",避免线性因果

五、数据校验(参照"对")

引用数字前先校验:绝对值/相对值、数量级、历史位置

六、信号与噪音

类型特征处理
真信号多源印证+具体细节直接引用
噪音单一来源+情绪化标题存疑,标注"待证实"
背景噪音已知事实重复简略提及

七、个人行动闭环

短期(1-3月):影响 → 建议
中期(6-12月):影响 → 建议
行动清单:[ ] 具体可执行的动作
三问:我能影响什么?我能准备什么?我能放弃幻想什么?

回答格式

## [主题] 分析(截至[日期])

### 一、时代背景(完整模式)
### 二、AI/技术变量(完整模式)
### 三、为什么是现在
### 四、跨事件联动
### 五、数据校验
### 六、信号与噪音
### 七、对个人的影响 + 行动清单

---
来源:...

禁止事项

  • ❌ 不做时代背景锚定(完整模式下)
  • ❌ 不确定的内容说"确定"
  • ❌ 不标注来源就下结论
  • ❌ 只报事件不找关联
  • ❌ 不考虑AI/技术变量
  • ❌ 引用过时信息不标注日期

参考资料

  • 关键市场比率(金油比/金房比/股金比)→ 按需从 references/market_ratios.md 读取
  • 完整分析示例 → references/example_analysis.md
  • 本技能可与 agent-reach 联动做多平台全网搜索

Comments

Loading comments...