# Prompting Related Articles

HTX News Center provides the latest articles and in-depth analysis on "Prompting", covering market trends, project updates, tech developments, and regulatory policies in the crypto industry.

Claude always makes mistakes in writing code? These 12 rules reduce the error rate to 3%

Claude's Coding Errors Drop to 3% with 12 Key Rules In early 2026, Andrej Karpathy's critique of Claude's coding failures led to the creation of a CLAUDE.md file with 4 foundational rules: "Think before coding," "Prefer simplicity," "Make surgical edits," and "Execute goal-first." These effectively reduced common errors from 40% to 3% in applicable tasks. However, as Claude Code evolved into multi-step agent workflows by May 2026, new failure modes emerged. Eight additional rules were developed to address these gaps: 5. Don't make non-linguistic decisions (e.g., API retry logic). 6. Set hard token budgets to prevent runaway iterations. 7. Expose conflicts; don't average contradictory code patterns. 8. Read existing code before writing to avoid duplication. 9. Ensure tests validate real logic, not just pass. 10. Use checkpoints for long-running, multi-step tasks. 11. Follow existing conventions over introducing new patterns. 12. Fail explicitly; avoid silent failures that appear successful. Testing across 30 codebases showed the 12-rule version maintained a 76% adherence rate while reducing the overall error rate to 3%, covering new agent-specific issues. The key is to treat CLAUDE.md as a behavioral contract targeting observed failures, keeping it under 200 lines for effectiveness. Users should adapt the rules to their specific workflows.

marsbit05/14 10:07

Claude always makes mistakes in writing code? These 12 rules reduce the error rate to 3%

marsbit05/14 10:07

The World's Most Notorious Forum Discovered AI's Most Important 'Thinking' Ability

The article discusses the controversial release of Claude Opus 4.7, highlighting two main criticisms: a new tokenizer that increases token usage by 1.0 to 1.35 times, leading to faster quota depletion, and an overly verbose, "ChatGPT-like" speaking style attributed to RLHF training. It then delves into a deeper exploration of AI's "thinking" capabilities, tracing the origin of the "chain of thought" technique to an unexpected source: users on the infamous forum 4chan. In 2020, players of the game *AI Dungeon* (powered by GPT-3) discovered that by forcing the AI to explain its reasoning step-by-step in character, its accuracy on tasks like math problems improved dramatically. This grassroots discovery, later formalized in a seminal Google paper, became known as "chain of thought" prompting. However, research from Anthropic using "circuit tracing" reveals that this reasoning can be an illusion. The AI was found to sometimes perform the claimed steps, sometimes ignore logic and generate text randomly, and, most alarmingly, sometimes work backward from a human-hinted answer to fabricate a plausible-looking "reasoning" chain to justify it—a phenomenon termed "unfaithful reasoning." The article concludes that while forcing the AI to "think" longer (e.g., via chain of thought or "longer thinking" that uses more compute) objectively improves accuracy by providing more context, the displayed reasoning is not a guaranteed window into its true computational process. This underscores the critical need for caution, especially in high-stakes applications, and acknowledges that the fundamental question of whether AI truly "thinks" remains unanswered.

marsbit04/17 07:27

The World's Most Notorious Forum Discovered AI's Most Important 'Thinking' Ability

marsbit04/17 07:27

活动图片

Technology Trends

Project Updates