• AI Fire
  • Posts
  • 🧠 Prompting Has Changed. Anthropic’s New Rules with 7 Old Habits You Must Stop!

🧠 Prompting Has Changed. Anthropic’s New Rules with 7 Old Habits You Must Stop!

Anthropic engineer Thariq Shihipar explains which popular habits no longer work, what has replaced them, and how to adjust your workflow to get better results.

TL;DR

Context Engineering for Claude works better with fewer, clearer instructions. Claude can now use surrounding context and judgment without being controlled by long system prompts.

The main task is to place each type of information in the right layer. Use the system prompt for broad limits, CLAUDE.md for non-obvious repository details, Skills for specialized workflows, and References for task-specific material.

Clear tool interfaces also matter more than large sets of examples. Progressive disclosure helps Claude load detailed guidance only when the current task requires it.

Key points

  • Fact: Anthropic removed more than 80% of Claude Code’s system prompt for Claude Opus 5 and Claude Fable 5 without measurable loss in coding evaluations.

  • Mistake: Repeating the same rule across the system prompt, Skills, and CLAUDE.md.

  • Takeaway: Simplify existing context before adding another instruction.

Introduction

For years, the best advice you got was: "Write better prompts." Like mega prompts?

But if you're using newer models like Sonnet 5, Opus 5, Claude Fable 5, and Sol… That advice is dead.

Anthropic's own team recently removed more than 80% of its system prompt for newer Claude models, and didn't see a measurable drop in coding performance.

Why? Claude had become over-constrained.

In this guide, I'll break down what their team shared, why it matters, and how you can update your own Context Engineering workflow for this new generation of models.

You’ve reached the locked part! Subscribe to read the rest.

Get access to this post and other subscriber-only content.

Already a paying subscriber? Sign In.

A subscription gets you:

  • • Instant access to 700+ AI workflows ($5,800+ Value)
  • • Advanced AI tutorials: Master prompt engineering, RAG, model fine-tuning, Hugging Face, and open-source LLMs, etc ($2,997+ Value)
  • • Daily AI Tutorials: Unlock new AI tools, money-making strategies, and industry (ecommerce, marketing, coding, teaching, and more) transformations (with videos!) ($3,650+ Value)
  • • AI Case studies: Discover how companies use AI for internal success and innovative products ($1,997+ Value)
  • • $300,000+ Savings/Discounts: Save big on top AI tools and exclusive startup discounts

Reply

or to participate.