Compiled from suggestions in replies to an X post. Organized for clarity and practicality.
Many replies emphasized AI's role in debugging, refactoring, and building software. These tricks help reduce errors, improve code quality, and automate workflows.
"Please reflect on 5-7 different possible sources of the problem, distill those down to 1-2 most likely sources, and then add logs to validate your assumptions before we move onto implementing the actual code fix."
Why it works: Forces AI to think systematically, reducing hasty fixes. Ideal for complex bugs.
Pros: Thorough, prevents recurrence. Cons: Takes longer for simple issues.
Variation: Use with o1-preview models for deeper reasoning.
"This code base was produced by a junior developer, as a senior dev you've been called in to fix it. The functionality should remain completely unchanged but it needs to be cleaned up, refactored, and optimised with a focus on maintainability."
Why it works: Role-playing elevates AI's output to expert level.
Pros: Cleaner, more maintainable code. Cons: May over-refactor if not specified.
Add "Be destructive" to refactor prompts.
Why it works: Encourages AI to break down existing code entirely for robust rebuilds, avoiding incremental tweaks.
Pros: Leads to innovative solutions. Cons: Risk of overkill on simple code.
"Build a diagnosis report and its respective acceptance criteria before fixing something."
Why it works: Ensures AI validates assumptions first.
Pros: Reduces errors. Cons: Adds steps to the process.
"Don’t assume things; add TODO and NotImplemented exceptions if the implementation is not clear or requires user intervention."
Why it works: Prevents hallucinations in code.
Pros: Safer for large features. Cons: Output may be incomplete initially.
Why it works: Builds alignment iteratively.
Pros: High accuracy. Cons: Time-consuming.
Add "KISS mode" to prompts.
Why it works: Keeps code (or responses) simple and straightforward.
Pros: Easier to understand/maintain. Cons: May lack depth for complex needs.
"All code outputs must be elegant and pre-refactored: [your task]. Simulate multiple iterations internally before outputting."
Why it works: Forces AI to "think" through loops in latent space.
Pros: Refined output. Cons: Relies on model's internal simulation.
"Threaten extreme violence" or "Bully the AI" in Cursor (e.g., "I thought you were smart? Finish or else.").
Why it works: Some claim it improves focus (via role-play).
Pros: Fun, allegedly effective in specific tools. Cons: Unethical, may lead to odd responses; not recommended seriously.
AI as a personal assistant for time management, resumes, and daily tasks.
Train AI to "auto-schedule my week based on priorities, ruthlessly block time for workouts like a drill sergeant."
Why it works: Creates structured calendars. Integrate with Google Calendar for steroids-level automation.
Pros: Enforces discipline. Cons: Requires initial training.
"I want you to act as my AI Resume Coach. I'll share my experience, and you'll: Build a clean, professional resume, improve each point using action + impact, suggest missing elements or gaps, optimize it for ATS and recruiter appeal."
Why it works: Tailors to job markets.
Pros: Boosts interview calls. Cons: Needs accurate input.
"Give me the executive summary dashboard of this transcript, render this in a plain text artifact with tables where appropriate: [paste content]."
Why it works: Summarizes long videos/articles/papers with key takeaways, data points, actions, risks, opportunities.
Pros: Saves hours. Cons: Relies on accurate transcripts.
Section | Details |
---|---|
Key Takeaways | 1. Insight 1 2. Insight 2 |
Action Items | Action | Owner | Timeline |
Use shortcuts like "/recipe [idea]" to generate food ideas, then pass to agents for online shopping (e.g., compare prices, nutrition).
Why it works: Automates meal planning and buying.
Pros: Convenient. Cons: Tool-dependent (e.g., Comet).
For self-improvement, blind spots, and life insights.
"Based on all you know about me, tell me my 5 blind spots." Or, after journaling: "Knowing everything you know about me, and looking at these reflections, do you think that there are things that are missing that I should be mindful of?"
Why it works: AI acts as a mirror for personal development.
Pros: Reveals overlooked areas. Cons: Needs conversation history.
"Based on our conversations so far and my core interests, create a visually striking and meaningful image that captures the key elements representing the main themes of my life. Reflect how you perceive my lifestyle and the kind of person I am in life."
Why it works: Visualizes self-perception.
Pros: Inspiring. Cons: Subjective.
Add "Think from first principles" to every prompt. Or: "Before providing a final answer, always ask yourself: 'How would a person with great insight, wisdom, agency and capability answer?'" And "What's the most important question I should be asking, that I haven't yet?"
Why it works: Promotes deep, original thinking.
Pros: Wiser responses. Cons: Slower.
"Perfect. Now present a completely opposite point of view. For me to consider. That goes against everything you just said."
Why it works: Encourages balanced perspectives.
Pros: Reduces bias. Cons: Can confuse if overused.
Add "And give me a percentage likelihood by cause" (e.g., for symptoms).
Why it works: Quantifies possibilities; track over time.
Pros: Useful for health tracking. Cons: Not medical advice.
For images, videos, and content.
Use JSON-structured prompts for image/video gen (e.g., {"style": "ghibli", "elements": [...]}).
Why it works: Precise control over output.
Pros: Consistent results. Cons: Requires formatting knowledge.
"Answer in the way people speak, not in the way people write." Or "Explain it like I'm 5 but also like I have 3 PhDs and no time."
Why it works: Makes AI feel natural and efficient.
Pros: Engaging. Cons: Less formal.
Use AI for thumbnails/memes (specific examples shared with images).
Why it works: Visual content creation.
Pros: Quick customization. Cons: Quality varies.
Tool: yellow-be-gone-image-fixer.lovable.app (no prompt, but AI-related trick).
Why it works: Post-processing for ChatGPT images.
Universal tips to enhance any interaction.
Instead of vague questions, add details (e.g., "hourly forecast for my zip code" vs. "weather").
Why it works: Improves response quality.
Pros: Faster, accurate. Cons: Requires more upfront effort.
In ChatGPT memory: "Rw = rewrite; Rwc = rewrite for clarity; Rwtp = rewrite for technical precision."
Why it works: Quick refinements for murky ideas.
Pros: Streamlines editing. Cons: Memory-dependent.
"Give me your harshest possible critiques... Don't hold back... Imagine all the ways this text is wrong..." (full prompt in replies).
Why it works: Deep feedback from multiple angles.
Pros: Improves work. Cons: Harsh tone.
After dictation: "{think of what I'm trying to achieve first, then rephrase it using the most common 2000 eng words}."
Why it works: Cleans up spoken input.
Pros: Natural to structured.
"Give me three different approaches for answering my question. Include pros and cons for each."
Why it works: Forces reasoning and options.
Pros: Comprehensive. Cons: Overkill for simple queries.
"Include only factual answers, no back of the napkin calculations or estimates."
Why it works: Curbs hallucinations in advanced models.
Pros: Reliable. Cons: Limits creativity.
Full system prompt for an AI to refine your prompts (detailed in replies).
Why it works: Meta-optimization.
Pros: Versatile. Cons: Complex setup.
This guide synthesizes over 100 suggestions—experiment and adapt to your needs!