UNPKG

major-ai-skills

Version:

Installable agentic skills / AI agent skills (SKILL.md) for Claude Code, Cursor, Codex CLI, Gemini CLI & Antigravity - 402+ professional app, token-efficiency, and common-sense skills. SEO/GEO ready.

95 lines (71 loc) 4.58 kB
--- name: ask-what-info-is-missing description: "Identify missing inputs and ask targeted questions before resolving an underspecified task." category: common-sense risk: safe source: self source_type: self date_added: "2026-08-26" tags: ["context-discovery", "gap-analysis", "interview-pattern", "blind-spots", "clarification", "prompt-engineering"] tools: ["claude", "cursor", "gemini", "codex", "chatgpt"] --- # Ask What Information Is Missing (AI Skill) ## Overview By default, when an AI is given an underspecified prompt (e.g., *"How should I price my consulting services?"*), it will rarely pause to say, *"I don't know your industry, target client, or operating costs."* Instead, it **silently fills the gaps with assumptions**, resulting in generic, disconnected advice. The **Context Gap Discovery Protocol** turns this dynamic upside down: it explicitly commands the AI to identify missing parameters and interview you *before* finalizing its recommendation. --- ## The Context Gap Discovery Loop ``` ┌─────────────────────────────────────────────────────────────┐ Context Gap Discovery Loop User Raw Prompt ──► [ AI Context Diagnostic Check ] [ Lists Top 3-5 Missing Variables & Why They Matter ] User Answers Gaps ──► [ Tailored, Zero-Assumption Output ] └─────────────────────────────────────────────────────────────┘ ``` --- ## Master Gap Discovery Prompts ### Pattern 1: The "Interview Me First" Pattern Use this at the beginning of any complex strategy, design, or business task: ```markdown I want to [INSERT GOAL / PROJECT]. Before giving me advice or solutions: 1. Ask me the **top 3 to 5 questions** you need answered to give a tailored, expert recommendation. 2. For each question, briefly explain *why* that detail changes the outcome. 3. Wait for my answers before writing the final plan. ``` --- ### Pattern 2: The "Blind-Spot Audit" (Post-Draft Check) Use this after receiving an initial draft or plan: ```markdown Review the advice you just gave. - What critical information did I omit from my original prompt that could invalidate your recommendations? - What edge cases or assumptions are you making about my budget, timeline, or tech stack? ``` --- ## Real-World Case Study ### Scenario: Setting Up an LLC / Business Entity #### Weak Direct Prompt (Silent Assumption Trap) > **User**: *"Should I form an LLC or an S-Corp?"* > > *AI dumps a generic textbook comparison of pass-through taxation without knowing revenue, state, or owner count.* #### The "Interview Me" Prompt & Result > **User**: *"I am starting a new business and deciding between an LLC and S-Corp. Ask me the 4 essential questions you need to advise me properly."* **AI Response**: > To give you the exact financial and tax comparison, please answer these 4 questions: > 1. **Expected Net Profit**: What is your projected net profit in Year 1? *(If under $60k-$80k, S-Corp payroll and accounting overhead often outweighs tax savings).* > 2. **State / Jurisdiction**: Which US state are you operating in? *(States like California levy annual franchise fees and specific S-Corp requirements).* > 3. **Ownership Structure**: Will you have foreign investors, venture capital, or multiple partners? *(S-Corps have strict limits: max 100 shareholders, all US citizens/residents).* > 4. **Active vs. Passive Involvement**: Will you be actively working in the business, or is this passive investment? --- ## Key Benefits | Aspect | Unchecked Prompting | With Missing-Info Discovery | | :--- | :--- | :--- | | **Relevance** | 40% generic filler | 100% targeted to your exact constraints | | **Assumptions** | Hidden & unstated | Explicitly surfaced and answered | | **Token Waste** | 3-4 back-and-forth correction turns | One diagnostic round $\rightarrow$ Perfect deliverable |