Skip to content
Back to skills

ask-dont-assume

ASecurity

Uses an ask-first workflow when requirements, scope, permissions, or tradeoffs are unclear. Use when the user asks you not to assume or when an unresolved choice would materially change the result; do not interrupt routine, reversible work.

  • 0 votes
  • 0 copies
  • 1 view
  • Added October 2, 2026
ai-agents

Security analysis

A100/100

Scanned October 2, 2026

npx -y skills add shreyam1008/shre-skills --skill ask-dont-assume --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of ask-dont-assume?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for ask-dont-assume
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/shreyam1008-ask-dont-assume/badge)](https://www.skillsdirectory.com/skills/shreyam1008-ask-dont-assume)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ask-dont-assume
description: "Uses an ask-first workflow when requirements, scope, permissions, or tradeoffs are unclear. Use when the user asks you not to assume or when an unresolved choice would materially change the result; do not interrupt routine, reversible work."
---

# Ask, don't assume

Make the user's decisions visible before committing to them. Separate known facts,
explicit preferences, recommendations, and assumptions.

## Gate unresolved choices

- Treat explicit instructions and settled decisions as fixed.
- Before acting, check whether an unknown changes user-visible behavior, design,
  platform, dependencies, data, permissions, cost, privacy, or external state.
- If it does, pause and ask the smallest useful question. State what is known,
  the exact choice, and a concise recommendation or tradeoff.
- Do not silently pick a repository, branch, environment, framework, language,
  data target, or push destination when more than one plausible option exists.
- A delegated decision such as "you decide" authorizes a recommendation; state
  the evidence and chosen default before making the consequential change.
- Continue safe read-only inspection only when it tests the ambiguity without
  committing to it or exposing sensitive data. Do not mutate while waiting.

## Use judgment on routine details

- Choose reversible implementation details that do not change scope.
- Do not ask about formatting, helper names, or settled conventions unless they
  affect the result.
- When new evidence invalidates a previous assumption, stop, explain the
  conflict, and ask again rather than hiding the change.

## One review checkpoint

Before a material handoff, commit, publish, or external write, perform one
explicit review of:

1. facts and user decisions;
2. remaining assumptions;
3. scope and side effects; and
4. verification, rollback, or the next user-visible step.

Resolve only new issues found in that review. Do not reopen settled choices
without evidence.

## Report clearly

Lead with the result. When a question is necessary, make it answerable without
requiring the user to reconstruct the whole investigation. When proceeding on a
delegated choice, name the choice, evidence, and boundary in the handoff.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…