Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.getsnippets.ai/llms.txt

Use this file to discover all available pages before exploring further.

The principle of specificity is the absolute bedrock of effective prompting. Vague prompts lead to vague, generic, and often useless answers. The more detail, clarity, and direction you can provide, the more the LLM can narrow its focus and deliver exactly what you need.

​
The Problem with Vague Prompts

When you give a model a vague prompt like “Tell me about dogs,” you’re forcing it to make a huge number of assumptions. It has to guess:
  • Topic: Are you interested in dog breeds, dog training, the history of dog domestication, or famous dogs in movies?
  • Depth: Do you want a single paragraph or a 2,000-word essay?
  • Audience: Is this for a child, a veterinarian, or a potential first-time dog owner?
  • Format: Should the output be a list, an article, or a poem?
The model will make a guess, but it’s unlikely to be the one you wanted.

​
From Vague to Hyper-Specific: A Case Study

Let’s look at how we can iteratively refine a vague prompt into a powerful and specific one. Vague Prompt: 
Write a story.
Slightly Better Prompt: 
Write a story about a detective.
This is better, but still very broad. What kind of detective? What kind of story? Good, Specific Prompt: 
Write a short story (around 500 words) about a hardboiled detective in 1940s New York City. The tone should be noir, with a sense of mystery and danger.
This is much better. It specifies the genre, setting, tone, and length. Excellent, Hyper-Specific Prompt: 
Write a 500-word short story in the style of Raymond Chandler. The protagonist is a cynical private investigator named Jack Corrigan, working in his dimly lit office in 1940s Manhattan. A mysterious woman in a red dress walks in, asking him to find her missing brother. The story should end on a cliffhanger.
This prompt gives the model everything it needs: a specific style to emulate, a character, a setting, a plot hook, and a structural constraint (the cliffhanger ending).

​
Actionable Techniques for Achieving Specificity

Here are key techniques you should practice to make your prompts laser-focused.
  • Specify the Desired Length: Don’t just say “short” or “long.” Give a word count (“about 200 words”), a sentence count (“in three sentences”), or a paragraph count (“in two paragraphs”).
  • Define the Output Format: Be explicit about the structure of the response.
    • “Provide the answer as a JSON object with the keys ‘name’, ‘capital’, and ‘population’.”
    • “Create a Markdown table with three columns: ‘Feature’, ‘Benefit’, and ‘Example’.”
    • “Write a numbered list of the top 5 action items.”
  • Set the Tone and Style: How should the AI sound?
    • “Use a formal, academic tone.”
    • “Write in a friendly, conversational, and encouraging style.”
    • “The tone should be witty and slightly sarcastic.”
  • Provide Negative Constraints: Sometimes it’s just as important to tell the model what not to do.
    • “Explain the concept of quantum physics without using any math.”
    • “Write a product description, but do not mention the price.”
    • “Summarize the article, excluding any information about the author’s personal life.”
By mastering the art of specificity, you move from getting random, hit-or-miss results to consistently getting the exact output you envision.