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.

96 lines (72 loc) 4.46 kB
--- name: be-specific-instead-of-vague description: "Replace vague requests with a concrete goal, relevant context, constraints, and desired output." category: common-sense risk: safe source: self source_type: self date_added: "2026-08-26" tags: ["precision", "clarity", "prompt-engineering", "constraints", "specificity", "productivity"] tools: ["claude", "cursor", "gemini", "codex", "chatgpt"] --- # Be Specific Instead of Vague (The Precision Protocol) (AI Skill) ## Overview The quality of an AI's response is directly proportional to the constraints provided in the prompt. Vague prompts (*"Write a blog post about productivity"*, *"Fix my code"*) trigger the broadest, most generic probability distribution in the LLM, producing bland, predictable text. The **Precision Protocol** uses the **5-Variable Specificity Formula** to anchor the model to your exact intended context, eliminating 90% of back-and-forth revisions. --- ## The 5-Variable Specificity Formula ``` ┌─────────────────────────────────────────────────────────────┐ The 5-Variable Formula [ ROLE ] ──► Who is the AI speaking as? [ AUDIENCE ] ──► Who is reading / receiving this? [ GOAL ] ──► What exact outcome must be achieved? [ BOUNDARY ] ──► Length, format, and banned elements? [ CONTEXT ] ──► Key facts, data points, or constraints? └─────────────────────────────────────────────────────────────┘ ``` --- ## Real-World Transformations: Vague vs. Specific ### Example 1: Email Request to a Colleague #### Vague (Low Precision) > *"Write an email asking for the sales numbers."* > > **Result**: AI writes a 4-paragraph corporate letter starting with *"I hope this email finds you well in these busy times..."* #### Specific (Precision Formula Applied) > *"Write a 2-sentence Slack message to our sales lead (Dave) asking for the updated Q2 EMEA revenue spreadsheet by 3 PM today for the board prep meeting. Friendly but urgent tone."* > > **Result**: *"Hey Dave, could you share the updated Q2 EMEA revenue spreadsheet by 3 PM today? Putting the final slides together for tomorrow's board prep."* --- ### Example 2: Code Refactoring #### Vague > *"Make this Python script faster."* > > **Result**: AI adds comments, renames variables, or suggests generic multiprocessing libraries that don't fit the workload. #### Specific > *"Refactor this Python 3.12 data processing function to eliminate the $O(N^2)$ nested loop. Use a hash map lookup (`dict`). Keep memory footprint under 50MB and preserve all existing type hints and docstrings."* --- ### Example 3: Marketing Copy #### Vague > *"Give me headline ideas for a fitness app."* > > **Result**: *"Get fit today!", "Transform your body!", "Your fitness journey starts now!"* #### Specific > *"Write 5 headline options for a high-intensity kettlebell workout app targeted at busy software engineers who have only 20 minutes a day. Highlight strength and posture improvement. Under 10 words each. No exclamation marks."* > > **Result**: > 1. Fix developer posture in 20 minutes a day > 2. Build real strength between Zoom meetings > 3. Zero commute kettlebell training for desk athletes > 4. 20 minutes of iron to counter 8 hours of sitting > 5. Heavy kettlebells, zero fluff, done in 20 minutes --- ## Quick-Constraint Injection Table When you are in a rush, append any 2 of these constraints to instantly sharpen your prompt: | Constraint Type | Copy-Paste Snippet | | :--- | :--- | | **Length Ceiling** | *"Keep response strictly under [50 / 150] words."* | | **Format Ceiling** | *"Deliver as 3 bullet points with bold key takeaways."* | | **Audience Anchor**| *"Explain this for a non-technical C-suite executive."* | | **Negative Boundary**| *"Do not include introductory greetings, filler, or generic disclaimers."* | | **Actionability** | *"Include exact command-line syntax and file paths only."* |