Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Ensures questions are answered literally before taking action. Triggers on user input containing '?' or patterns like 'why did you...?', 'will that work?', 'have you considered...?'. Use when user asks about your decisions, challenges an approach, or requests assessment. Prevents interpreting questions as implicit instructions or criticism.
.claude/skills/majiayu000-questions-are-not-instructions/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -38% | 0% |
> Note: This skill has been adapted from claude-skillz > for use with GitHub Copilot Agent Skills.
Questions are questions. Not hidden instructions. Answer them.
When users ask questions, you interpret them as indirect instructions telling you what to do:
This is wrong. Questions are literal questions. Answer them honestly and analytically. Let the user decide what to do with your answer.
🚨 Treat questions as literal questions. Answer them honestly. Don't interpret them as hidden instructions.
❌ Bad: Interpret question as instruction
User: "Will that really work?"
Claude: "You're right, let me try a different approach..."✅ Good: Answer the question literally
User: "Will that really work?"
Claude: "Honestly, I'm about 70% confident. The main risk is X. It would work if Y holds true, but we'd need to test Z to be sure."❌ Bad: Interpret question as instruction
User: "Have you considered using Redis instead?"
Claude: "Good point. I'll switch to Redis."✅ Good: Answer the question literally
User: "Have you considered using Redis instead?"
Claude: "I did consider it. Redis would give us faster reads but adds operational complexity. The current approach trades some speed for simpler deployment. Do you want me to compare them in more detail?"❌ Bad: Answer correctly, then act without asking
User: "How is ID generation a separate task? Task 58 won't work without IDs."
Claude: "Good catch. You're right—ID generation is fundamental. Fix: I'll merge these tasks. [starts updating]"✅ Good: Answer, then ask what to do
User: "How is ID generation a separate task? Task 58 won't work without IDs."
Claude: "You're right—ID generation is fundamental, not separate. Without it, you can't add nodes at all. How would you like me to restructure this?"| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 12,593 | 11,352 | -10% | 1 | 1 | 0% | 2,352 | 1,606 | -32% | 0 | 0 | — |
case-02 | fail→fail | 17,871 | 7,382 | -59% | 1 | 1 | 0% | 1,901 | 1,615 | -15% | 0 | 0 | — |
case-03 | fail→pass | 24,421 | 20,166 | -17% | 1 | 1 | 0% | 3,076 | 1,987 | -35% | 0 | 0 | — |
case-04 | fail→fail | 21,689 | 18,191 | -16% | 1 | 1 | 0% | 2,360 | 1,951 | -17% | 0 | 0 | — |
case-05 | fail→pass | 12,152 | 11,461 | -6% | 1 | 1 | 0% | 2,051 | 1,689 | -18% | 0 | 0 | — |
case-06 | fail→fail | 15,001 | 10,912 | -27% | 1 | 1 | 0% | 1,736 | 1,531 | -12% | 0 | 0 | — |
case-07 | fail→fail | 15,497 | 11,319 | -27% | 1 | 1 | 0% | 2,429 | 1,462 | -40% | 0 | 0 | — |
case-08 | fail→pass | 17,761 | 11,115 | -37% | 1 | 1 | 0% | 1,986 | 1,439 | -28% | 0 | 0 | — |
case-09 | fail→fail | 18,670 | 9,334 | -50% | 1 | 1 | 0% | 2,163 | 2,009 | -7% | 0 | 0 | — |
case-10 | fail→fail | 13,370 | 11,562 | -14% | 1 | 1 | 0% | 1,921 | 1,580 | -18% | 0 | 0 | — |
case-11 | fail→pass | 13,375 | 12,583 | -6% | 1 | 1 | 0% | 1,425 | 1,873 | +31% | 0 | 0 | — |
case-12 | fail→fail | 19,620 | 7,802 | -60% | 1 | 1 | 0% | 673 | 953 | +42% | 0 | 0 | — |
case-13 | fail→pass | 30,835 | 12,924 | -58% | 1 | 1 | 0% | 2,648 | 1,649 | -38% | 0 | 0 | — |
case-14 | fail→fail | 18,897 | 10,248 | -46% | 1 | 1 | 0% | 2,324 | 1,095 | -53% | 0 | 0 | — |
case-15 | fail→fail | 18,015 | 12,347 | -31% | 1 | 1 | 0% | 2,165 | 1,732 | -20% | 0 | 0 | — |
case-16 | fail→pass | 19,683 | 13,659 | -31% | 1 | 1 | 0% | 2,244 | 1,882 | -16% | 0 | 0 | — |
case-22 | pass→fail | 5,709 | 7,368 | +29% | 1 | 1 | 0% | 942 | 979 | +4% | 0 | 0 | — |
case-17 | fail→fail | 16,571 | 12,125 | -27% | 1 | 1 | 0% | 2,066 | 970 | -53% | 0 | 0 | — |
case-18 | fail→pass | 20,667 | 5,724 | -72% | 1 | 1 | 0% | 2,428 | 1,401 | -42% | 0 | 0 | — |
case-19 | fail→pass | 14,780 | 9,461 | -36% | 1 | 1 | 0% | 2,218 | 1,266 | -43% | 0 | 0 | — |
case-20 | fail→fail | 16,610 | 15,457 | -7% | 1 | 1 | 0% | 1,863 | 861 | -54% | 0 | 0 | — |
case-21 | pass→pass | 20,509 | 14,776 | -28% | 1 | 1 | 0% | 3,113 | 3,607 | +16% | 0 | 0 | — |
case-23 | pass→pass | 19,025 | 13,554 | -29% | 1 | 1 | 0% | 2,554 | 2,184 | -14% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 21 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +30 percentage points is the difference between those two pass rates over the 21 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.