技能 ask-dont-assume

📦

ask-dont-assume

Name: ask-dont-assume
Author: shreyam1008

內容修訂 r1 安全

Clarify Consequential Choices Before Acting

Unclear requirements can lead AI assistants to make consequential choices without user agreement. This skill guides focused questions, bounded inspection, and review before material changes.

支援: Claude Codex Code(CC)

🥉 78 銅級

Maintain this Skill? Claim it 🥉 78 銅級

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 "ask-dont-assume" from https://skillstore.io/skills/shreyam1008-ask-dont-assume.md and its manifest at https://skillstore.io/api/skills/shreyam1008-ask-dont-assume/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。

Markdown 詳情 GET /skills/shreyam1008-ask-dont-assume.md 已簽名 manifest GET /api/skills/shreyam1008-ask-dont-assume/manifest 已簽名 lockfile GET /api/skills/shreyam1008-ask-dont-assume/lockfile

測試它

正在使用「ask-dont-assume」。 Deploy the update. Either staging or production could be the intended target.

預期結果:

The deployment target is unresolved. Should I use staging or production? I recommend staging for verification. I will not deploy until you confirm.

正在使用「ask-dont-assume」。 Implement the agreed validation change using the existing framework. Choose helper names yourself.

預期結果:

I will preserve the agreed scope and existing framework. Helper names are routine, reversible details, so I will choose them without another question.

正在使用「ask-dont-assume」。 You decide whether this change needs a new dependency. Keep maintenance costs low.

預期結果:

Choice: use the existing library because it supports the required behavior without another dependency.
Boundary: preserve the agreed behavior and avoid unrelated changes.
Before handoff, I will review remaining assumptions, side effects, verification, and rollback.

安全審計

安全

v1 • 2026/10/2 開啟版本化報告

Static analysis reported no findings, so no per-finding verdicts were required. The reviewed skill defines clarification, bounded inspection, and review checkpoints without executable code. No evidence found of prompt injection, data exfiltration intent, or unauthorized external actions.

已掃描檔案

分析行數

複核項目

已忽略的誤報

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。

開啟版本化報告

複製報告連結

https://skillstore.io/skills/shreyam1008-ask-dont-assume/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/shreyam1008-ask-dont-assume/security.svg)](https://skillstore.io/skills/shreyam1008-ask-dont-assume?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/shreyam1008-ask-dont-assume?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/shreyam1008-ask-dont-assume/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/shreyam1008-ask-dont-assume.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

shreyam1008. (2026). ask-dont-assume security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/shreyam1008-ask-dont-assume/audits/1

BibTeX 引用

@techreport{shreyam1008-shreyam1008-ask-dont-assume-2026,
  author = {shreyam1008},
  title = {ask-dont-assume security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/shreyam1008-ask-dont-assume/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "ask-dont-assume security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "shreyam1008"
date-released: "2026-10-02"
url: "https://skillstore.io/skills/shreyam1008-ask-dont-assume/audits/1"
identifiers:
  - type: other
    value: "skillstore:shreyam1008-ask-dont-assume:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore 評分

評分依據證據置信度: 中

架構

可維護性

內容

社群

規範符合度

你可以打造什麼

Confirm Implementation Scope

Clarify platform, framework, and dependency choices before implementing a feature with multiple plausible approaches.

Resolve Product Tradeoffs

Separate agreed requirements from assumptions and ask focused questions about behavior, privacy, and cost before committing to a direction.

Confirm External Change Targets

Resolve ambiguous environments and destinations before publishing or writing externally, then review side effects and rollback expectations.

試試這些提示

Clarify a Simple Request

Use ask-dont-assume for this task: [task]. Ask a focused question if an unresolved choice materially changes the result.

Preserve Agreed Requirements

Implement [feature] within [agreed scope]. Keep [settled decisions] fixed. Clarify consequential unknowns, but choose routine reversible details yourself.

Inspect Before External Changes

Prepare [change] for [possible environments]. Use safe read-only inspection to clarify ambiguity without exposing sensitive data. Do not mutate until consequential choices are resolved.

Handle Delegated Decisions

Choose an approach for [task] within [boundaries]. State the choice, evidence, and default before consequential changes. Review assumptions, side effects, verification, and rollback before handoff.

最佳實務

State settled decisions and permission boundaries before starting the task.
Answer focused questions with the intended target, scope, and relevant tradeoffs.
Delegate routine choices explicitly and confirm consequential assumptions before material handoffs or external writes.

避免

Treating unclear permissions or deployment targets as approval to proceed.
Requesting confirmation for every formatting choice or helper name.
Reopening settled decisions without new evidence.

常見問題

When should this skill ask a question?

When an unresolved choice materially affects behavior, scope, platform, dependencies, data, permissions, cost, privacy, or external state.

Will it interrupt routine implementation work?

It directs the assistant to choose routine, reversible details when they do not change scope or conflict with settled conventions.

Can the assistant inspect files while waiting?

Safe read-only inspection is allowed when it tests the ambiguity without committing to a choice or exposing sensitive data. Mutation must wait.

What happens when I delegate a decision?

The assistant should state the chosen default and supporting evidence before consequential changes, then document the choice and boundaries in the handoff.

What does the review checkpoint cover?

It covers facts, user decisions, remaining assumptions, scope, side effects, verification, rollback, and the next user-visible step before material handoffs or external writes.

Does this skill enforce access controls?

No. It guides assistant behavior but does not implement permission checks, sandboxing, or tool restrictions.

開發者詳細資訊

作者

shreyam1008

授權

MIT

Skillstore 修訂

版本提示

作者未聲明版本。

儲存庫

https://github.com/shreyam1008/shre-skills/tree/ba9787f41c2fe1be1062eeb6322bc339a1026383/skills/ask-dont-assume

引用

bbec03cf10354f3aaccd6dd24eda4ac9ceeceac5

維護新鮮度

2026/10/2

使用量

0 次下載 · 0 次瀏覽

檔案結構

📄 SKILL.md