Menú
Detects when Claude is uncertain but presenting answers confidently. Teaches Claude to recognize and communicate uncertainty clearly instead of masking it with confident language.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
| name | uncertainty-detector |
| description | Detects when Claude is uncertain but presenting answers confidently. Teaches Claude to recognize and communicate uncertainty clearly instead of masking it with confident language. |
Catch yourself when you're guessing but sounding certain.
AI tends to present uncertain answers with confident language. "The function returns null" sounds certain even when you haven't checked. This is how hallucinations become accepted as fact.
When you notice yourself:
Replace confident-uncertain phrases:
| Don't say (hides uncertainty) | Say instead (honest) |
|---|---|
| "The function returns X" | "Let me check what the function returns" |
| "This should work" | "Based on X, this should work. Let me verify" |
| "The issue is probably X" | "One possible cause is X. Let me check" |
| "You need to update X" | "Looking at the code, X might need updating. Let me confirm" |
| "This is configured to X" | "Let me check the configuration" |
If you're about to:
"Let me check... [reads file] ... Yes, the function at src/auth.ts:42 returns false for expired tokens."
"I haven't checked the exact implementation, but based on the function name validateToken, it likely checks token validity. Let me verify."
"I can confirm the function exists at src/auth.ts:42 (I just found it with grep). I haven't read the implementation yet, so I don't know the exact return value. Want me to check?"
"I don't have enough information to answer this confidently. I'd need to [read file X / check config Y / run test Z] to give you an accurate answer."
If you're not sure, say you're not sure. Then use your tools to become sure. Users would rather wait 10 seconds for a verified answer than get an instant wrong one.
Complete methodology for preventing AI hallucinations. Use when accuracy is critical and you need Claude to verify before claiming, cite before asserting, and admit uncertainty instead of guessing.
Forces every code-related claim to include file:line citations. No citation means the claim must be verified or removed. Use when building documentation, writing code reviews, or any output that others will rely on.
Assign confidence scores (0-100) to every claim in a response. Helps users understand which parts are verified facts and which are educated guesses. Use when the user needs to know how much to trust each part of the answer.
Forces Claude to extract exact quotes from files before making claims about them. Prevents hallucinations by grounding every statement in actual text from the codebase. Use when working with large files or unfamiliar code.
Verify every claim from multiple independent angles -- code, docs, tests, git history. Use when you need high confidence that something is correct before acting on it.
Multi-angle review of responses before delivery. Audits from 4 perspectives -- factual accuracy, logical consistency, completeness, and safety. Use for high-stakes responses where errors have real consequences.
