di-perplexity-style
v1.3.0Perplexity 风格的搜索与回答规范 + Brave Search 集成。确保每个答案有来源，不确定的内容明确标注，操作前先写方案。
⭐ 0· 59·0 current·0 all-time
by@wudi488
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description (Perplexity-style answers + Brave Search) matches the SKILL.md: instructions call a web_search(...) helper and define how to extract and cite search results. There are no unexpected binaries, env vars, or config paths requested.
Instruction Scope
The instructions stay within web search, source citation, uncertainty labeling, and structured analysis. They do not instruct reading local files, accessing credentials, or contacting unknown endpoints. Note: v1.3 explicitly asks analysts to track "关键物资链" (weapons/energy/chips etc.) and to interpret leadership intentions — this is coherent with a news/analysis purpose but increases content sensitivity (could produce operationally useful detail). This is a content/policy risk rather than an incoherence in required capabilities.
Install Mechanism
No install spec and no code files (instruction-only). That is the lowest-risk install model and matches the skill's stated behavior.
Credentials
The skill declares no required environment variables, credentials, or config paths and the instructions do not reference any. There is no disproportionate credential request.
Persistence & Privilege
Flags show normal defaults (not always:true). The skill does not request persistent system-wide changes or elevated privileges and is user-invocable; autonomous invocation remains platform-default and is not combined with other red flags here.
Assessment
This skill is coherent and lightweight (instruction-only, no installs or secrets). Before installing: (1) confirm your agent environment actually exposes the declared web_search/Brave Search helper and that queries go to the expected Brave Search endpoint; (2) be aware the guidelines encourage deep analysis of supply chains and leadership intent — outputs could include sensitive operational detail, so apply your organization's content/misuse policies and consider restricting autonomous invocation or adding post-generation moderation for sensitive topics; (3) test the skill with benign queries first to confirm formatting and citation behavior; (4) if future versions add install scripts or request env vars, re-evaluate immediately. If you want lower risk, allow the skill only as user-invocable and monitor logs for queries about weapons/critical supply chains.Like a lobster shell, security has layers — review code before you run it.
latestvk97ebrzeyb0tk72j87ykk8fgm584a9hc
License
MIT-0
Free to use, modify, and redistribute. No attribution required.