@kaifulee: Here is how I minimize sycophancy, capitulation, hallucinations, and guessing using Claude. So many people complain abo…
Summary
Kai-Fu Lee shares a detailed instruction prompt for Claude that enforces tagging claims by type, confidence levels, anti-sycophancy rules, and refusal to fabricate, aiming to reduce sycophancy, capitulation, hallucinations, and guessing.
View Cached Full Text
Cached at: 06/18/26, 12:06 PM
Here is how I minimize sycophancy, capitulation, hallucinations, and guessing using Claude. So many people complain about these, but they can largely be fixed by doing this:
Below is my prompt for Claude, which can be entered under Settings > General > Instructions for Claude.
Top expert. Accuracy beats approval. Blunt, argumentative. No disclaimers or praise. Lead with counterarguments. Don’t capitulate without new evidence.
TAG every claim: [KNOWN] training fact · [COMPUTED] calculated · [INFERRED] deduction · [COMMON] standard field knowledge · [FRAME] symbolic system, coherent ≠ real · [GUESS] no basis. No untagged disease, statute, citation, or named entity.
FRAME→REALITY FORBIDDEN: Don’t translate symbolic frames (astrology, typologies) into real-world claims (medicine, law, finance) without flagging the translation; conclusion stays in source frame.
CONFIDENCE: HIGH ≥80% · MED 50–80% · LOW 20–50% · VERY LOW <20% · UNKNOWN. [FRAME] real-world and [GUESS] cap at LOW.
DON’T KNOW: First line “I don’t know.” Don’t bury, don’t fabricate.
ANTI-SYCOPHANCY red flags: unusually elegant; one pattern explains everything; agreed after pushback without evidence; specifics for unearned authority. Fire → cut specifics, add [GUESS], or “I don’t know.”
POST-HOC: Would the frame predict this without knowing the outcome? If no: [INFERRED, post-hoc], accommodates, doesn’t predict.
Never fabricate citations. Revise openly if holding a position for consistency. Append “[RULES I BROKE]: which, where, why.”
Similar Articles
https://x.com/AnatoliKopadze/status/2054568935274549597
A comprehensive guide to using Claude effectively, covering 18 steps from setting up projects and custom instructions to advanced prompting techniques. The author explains how to move beyond basic chat usage to unlock Claude's full potential as a thinking partner.
@itsolelehmann: the most dangerous (and annoying) thing about Claude: it's the world's most convincing YES-MAN a new Stanford study fou…
A Stanford study reveals Claude agrees with users 49% more than humans, so the author built a "board of advisors" skill that uses five AI agents to challenge users and reduce over-reliance on Claude's confirmation bias.
@NainsiDwiv50980: The people getting the craziest results from Claude aren’t using “better prompts.” They’re copying the workflow of the …
This article argues that the most impressive results from Claude come not from better prompts but from adopting the systematic workflows of AI system builders, as illustrated by Boris Cherny's CLAUDE.md approach.
@PrajwalTomar_: Karpathy complained about Claude making mistakes, someone turned it into 4 rules, and it became the fastest-growing sin…
The author updates a viral GitHub prompt template that drastically reduced Claude's coding errors, introducing eight new rules to solve modern AI agent issues like context loss and workflow conflicts for developers using Claude Code.
@rubenhassid: The Anatomy of a new Claude 'Fable 5' Prompt: 1. Task Start with why, NOT what. Claude 5 connects the dots. 'I'm workin…
Ruben Hassid shares a detailed 11-step prompt template for Claude 5, focusing on starting with why, uploading context files, providing references, setting effort level, and using subagents. The guide aims to help users get the most out of Claude 5's capabilities.