技能 deepen
📦
deepen
v1.0.0 內容修訂 r1 安全 ⚙️ 外部命令🌐 網路存取
Deepen Tasks Into Evidence-Based Prompts
Vague prompts can omit primary sources, specific methods, and reliable checks. Deepen rewrites the same task with concrete evidence and validation while preserving its scope.
支援: Claude Codex Code(CC)
使用我的智慧代理安裝
將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。
智能體請求
Review the Skillstore skill "deepen" from https://skillstore.io/skills/elrazez-deepen.md and its manifest at https://skillstore.io/api/skills/elrazez-deepen/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。
Agent 可讀資源
當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。
測試它
正在使用「deepen」。 Deepen a request to explain a conversion-rate decline using supplied event data and metric definitions.
預期結果:
- Professional approach: check whether tracking or denominator changes explain the decline before suggesting causes.
- Primary material: use the supplied event records and metric definitions.
- Methods and tools: compare equivalent time windows using the available analysis tools.
- Independent check: reconcile calculated totals with the supplied reporting baseline.
- Rewritten prompt: Explain the conversion-rate decline using the supplied events and definitions. Check tracking consistency and comparable windows. Reconcile totals before stating conclusions.
正在使用「deepen」。 Deepen a request to revise an announcement using supplied release notes and an editorial checklist.
預期結果:
- Professional approach: separate supported product claims from assumptions before revising the announcement.
- Primary material: use the supplied release notes and editorial checklist.
- Methods and tools: trace each factual claim to the release notes and apply the checklist.
- Independent check: every product claim must have support, and every applicable checklist requirement must pass.
- Rewritten prompt: Revise the announcement using the supplied release notes. Preserve its audience and scope. Support product claims and satisfy the editorial checklist.
安全審計
安全v1 • 2026/10/16 開啟版本化報告
All five static findings are false positives: Markdown code spans, a homepage metadata link, and guidance against scope expansion. No evidence found of malicious instructions, executable shell payloads, credential collection, or data exfiltration in SKILL.md. Optional source reads support prompt refinement; the skill explicitly stops before executing the task.
1
已掃描檔案
58
分析行數
0
複核項目
0
已忽略的誤報
風險因素
⚙️ 外部命令 (3)
🌐 網路存取 (1)
最近一次完成的靜態與語意稽核未偵測到已確認的安全發現。這並不能證明該 Skill 沒有副作用。
審計單位: codex
分享與引用此報告
分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。
複製報告連結
https://skillstore.io/skills/elrazez-deepen/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_reportMarkdown 徽章
[](https://skillstore.io/skills/elrazez-deepen?utm_source=security_passport_badge)HTML 徽章
<a href="https://skillstore.io/skills/elrazez-deepen?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/elrazez-deepen/security.svg" alt="Skillstore security assessment" loading="lazy"></a>嵌入卡片
<iframe src="https://skillstore.io/embed/skills/elrazez-deepen.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>學術引用 (APA · BibTeX · CFF)
APA 引用
elrazez. (2026). deepen security audit report (audit version 1) [Author version 1.0.0]. Skillstore. https://skillstore.io/skills/elrazez-deepen/audits/1BibTeX 引用
@techreport{elrazez-elrazez-deepen-2026,
author = {elrazez},
title = {deepen security audit report (audit version 1)},
institution = {Skillstore},
year = {2026},
number = {1},
url = {https://skillstore.io/skills/elrazez-deepen/audits/1},
note = {Author version 1.0.0}
}CITATION.cff
cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "deepen security audit report (audit version 1)"
version: "1.0.0"
type: report
authors:
- name: "elrazez"
date-released: "2026-10-16"
url: "https://skillstore.io/skills/elrazez-deepen/audits/1"
identifiers:
- type: other
value: "skillstore:elrazez-deepen:audit:1"
description: "Skillstore immutable audit report identifier"
Skillstore 評分
評分依據 證據置信度: 中 55
架構
100
可維護性
87
內容
65
社群
83
規範符合度
你可以打造什麼
Prepare a focused engineering task
Refine a bug-fix request around relevant project sources, available tools, and a regression check without adding features.
Ground a research request
Identify primary evidence, suitable methods, and an external validation criterion before starting a research task.
Sharpen a writing brief
Clarify source material, editorial methods, and verifiable requirements while keeping the original writing deliverable.
試試這些提示
Refine a simple request
/deepen Write a project status update using the notes I provide. Keep the same deliverable and do not write the update yet.
Use project context
/deepen Fix the bug described below. Use the provided repository context to name relevant sources and a regression check without implementing the fix.
Improve a research method
/deepen Compare the two policies in the attached documents. Identify primary evidence, an appropriate method, and an independent check without performing the comparison.
Refine a constrained investigation
/deepen Investigate the latency regression described below using the supplied measurements. Name available tools and a pass-or-fail check; preserve scope and do not execute.
最佳實務
- Provide the original task, relevant context, and scope limits before requesting a rewrite.
- Supply accessible primary sources and name tools that are actually available.
- Review the proposed checks and approve the rewritten prompt before starting execution.
避免
- Requesting immediate task execution when only prompt refinement is intended.
- Adding new features or deliverables under the label of deeper analysis.
- Accepting generic advice or self-assessed quality instead of a concrete independent check.
常見問題
Does Deepen perform the task?
No. It rewrites the prompt and stops for the user before execution.
How do I invoke it?
Use /deepen followed by your task, $deepen in Codex, or ask for a deeper version of a prompt.
What does the response contain?
Four short answer lists followed by one fenced prompt block of at most 12 lines.
Does it require file or web access?
No. Without access, it uses your context and names sources that the rewritten prompt should consult.
Can it use project-specific guidance?
Yes. The procedure directs the assistant to read relevant local skill guidance or documents naming project sources, tools, and checks.
What happens when I provide no task?
It asks for the task in one line and stops.