Claude Secret Commands: 50 Shortcuts That Save Hours

Claude responds differently from ChatGPT. These shortcuts are tuned specifically for Claude's behaviour and strengths.

PRECISION COMMANDS (1-10)

1. "Follow my instructions literally" — Claude will stick to exactly what you asked without adding unrequested content
2. "Be direct. No filler." — Removes lengthy preambles and unnecessary caveats
3. "Challenge me if I'm wrong about something" — Claude is more willing than most AIs to push back when prompted
4. "Don't add suggestions I didn't ask for" — Stops Claude from expanding scope beyond the task
5. "If you're uncertain about anything in my request, ask before proceeding" — Prevents guessing and wasted output
6. "Think through this before responding" — Claude handles extended thinking well; use this for complex reasoning
7. "Give me your actual opinion, not a balanced summary" — Claude will take a position when explicitly asked
8. "What would you do if this were your problem?" — Gets more personal, actionable advice
9. "What am I not asking that I should be?" — Claude is good at surfacing blind spots
10. "Be the expert, not the assistant" — Shifts tone from helpful to authoritative

DOCUMENT SHORTCUTS (11-20)

These leverage Claude's long context window: 11. "Read everything I paste before responding" — Ensures Claude processes long inputs fully 12. "Summarise this for: an executive (2 min read), a team member (5 min read), a detail-oriented reader (10 min read)" — Three audiences, one prompt 13. "Find every instance where [THING] is mentioned and list them with page context" 14. "Extract all [DATA TYPE: dates/names/numbers/commitments] from this document" 15. "Compare document A and document B. What's different? What's missing from each?" 16. "What's the weakest argument in this document?" 17. "What assumptions is this document making that aren't stated?" 18. "Identify any contradictions in this document" 19. "What questions would a skeptic ask about the claims in this document?" 20. "Write an executive summary that captures the spirit, not just the facts"

WRITING SHORTCUTS (21-30)

21. "Match the tone of this example exactly: [PASTE EXAMPLE]"
22. "Write it at a [GRADE LEVEL] reading level"
23. "Make it sound like I wrote it, not an AI" — Claude is better at this than most models
24. "Give me the version that gets to the point in the first sentence"
25. "Write the opening paragraph 5 different ways"
26. "Remove every adjective that isn't doing critical work"
27. "Rewrite this so the most important thing is said first"
28. "What's the ideal length for this piece and why?"
29. "Where does the reader's attention drop and how do we fix it?"
30. "Write it so a smart 16-year-old would find it genuinely interesting"

ANALYSIS SHORTCUTS (31-40)

31. "What are the second-order consequences of [DECISION]?"
32. "What would this look like in 10 years if it goes perfectly? If it goes badly?"
33. "Where is the biggest risk in this plan that people typically overlook?"
34. "Give me the pre-mortem: this project failed. What went wrong?"
35. "What would need to be true for the opposite approach to be correct?"
36. "Stress-test this assumption: [ASSUMPTION]"
37. "What would a domain expert see that a generalist would miss?"
38. "Give me the bear case and the bull case for [SITUATION]"
39. "What's the minimum version of this that would still work?"
40. "Map out the dependencies: what has to happen before [THING] can happen?"

CODING SHORTCUTS (41-50)

These work especially well in Claude: 41. "Explain this code as if I've never seen the language before" — [PASTE CODE] 42. "Find every potential bug in this code. Be exhaustive." 43. "Rewrite this function to be more readable. Prioritise clarity over cleverness." 44. "Add error handling for every edge case you can identify" 45. "What would break this code? Give me 5 specific inputs that would cause problems." 46. "Write tests for this function that cover edge cases, not just the happy path" 47. "Is there a simpler way to achieve the same result?" 48. "Review this for security vulnerabilities. Be specific." 49. "What's the performance bottleneck and how would you fix it?" 50. "Convert this to [LANGUAGE]. Maintain the same logic but follow idiomatic [LANGUAGE] conventions."

Downloaded from DavideGPT.ai — Follow @DavideGPT for daily AI tips