7 Rules for Prompting Claude 5

Anthropic Just Revealed 7 New Rules for Prompting Claude 5 Models

🎬 Ben AI 📅 Sep 2, 2026 ⏱ 13:15
prompting Claude Opus 5 Fable 5 context engineering Anthropic

🎯 The Shift

Claude Opus 5 and Fable 5 work fundamentally differently from previous models — "almost everyone is prompting them wrong." They're trained for end-to-end task execution, which inverts the old prompting wisdom. 0:00

Boris Cherny (YC keynote): "A really common mistake is giving overly specific instructions — 'do it this way, then this, then this.' For modern models that's really not the way. Describe the task, the guardrails, the exit criteria — and let the model cook. This wouldn't have worked six months ago, but it does today."

🔧 All 7 Rules

1
Give the whole job, not step-by-step.

Claude 5 models perform best with the complete task spec up front and are left to run. The framework: job + why + guardrails + what "done" looks like. Ben used it to prepare the video itself — a broad goal + why + guardrails + done-definition, let it run 10–15 min, got a useful pre-outline.

2
Use the "interview me" skill before complex tasks.

For complex end-to-end tasks it's hard to articulate every guardrail upfront. Anthropic internally uses an "interview me" skill that asks you questions about the task, surfaces unknowns, then produces a brief/prompt. (Karpathy's alternative: voice-transcribe and ramble for a few minutes — then structure it.) "Put more effort into pre-planning instead of quick-prompt-and-iterate."

3
Prompt the WHY, not just the what.

On complex tasks the model hits decisions you didn't pre-specify. Understanding intent lets it make those small calls. The template: "I'm working on [larger task] for [specific person]. They need [what the output enables]. With that in mind, [request]." — same principle as delegating to a human.

4
Define what "done" looks like.

Claude 5 models are long-running — "they don't do too little, they do too much." Define exit criteria + output style (e.g. "8–15 tips with examples and sources" + a reference example of the format). Fix the Opus 5 verbosity: "give the output in plain English that's easy to understand."

5
Swap hard rules for reasons.

Instead of "never do X" / "avoid this," give the instruction with a reason. The new models respond far better to reasoned guardrails. (E.g. "make sure points are backed up by Anthropic's own team, so we have proof for our claims.") Anthropic even launched a rule rewriter skill to convert your CLAUDE.md/skills.

6
Avoid double-checking prompts.

Opus 5/Fable 5 already verify and fix mistakes autonomously — instructing "double-check" or using a sub-agent to verify adds cost without improving results. Also avoid: "think step-by-step," "explain your reasoning," and aggressive CAPS emphasis (causes over-triggering).

7
Fix Claude's voice once.

Opus 5 gets jargon-heavy/verbose. Anthropic recommends a simple prompt in CLAUDE.md, desktop global instructions, or system prompt — e.g. "keep responses focused, brief, and concise; avoid jargon and being overly verbose." Set it once, don't re-fight it every prompt.

🥇 The Golden Rule

Anthropic's own framing: "Think of Claude as a brilliant but new employee who lacks context on your norms and workflows." 12:45

"Show your prompt to a colleague with minimal context and ask them to follow it. If they'd be confused, Claude will be too." — We have to start prompting these models more and more like humans.

💡 Key Takeaways

  1. Prompt the whole job (job + why + guardrails + done), not step-by-step instructions.
  2. Pre-plan with the "interview me" skill (or voice-rambling) before complex tasks — reduces iteration and token burn.
  3. Give the why — intent drives better decisions on unspecified details.
  4. Define done + output style — long-running models over-produce without exit criteria.
  5. Reasoned guardrails beat hard rules — and the rule rewriter skill converts your CLAUDE.md.
  6. Stop adding verification — these models self-verify; avoid step-by-step and CAPS emphasis.
  7. Fix the voice once in global instructions, don't re-specify style every prompt.

📍 Timestamp Index

0:00 Intro
0:37 Rule 1 — whole job, not steps
3:00 Rule 2 — interview me skill
6:21 Rule 3 — prompt the why
7:57 Rule 4 — define done
9:17 Rule 5 — hard rules → reasons
10:12 Rule 6 — avoid double-checking
11:38 Rule 7 — fix the voice once
12:45 The golden rule
☰ View all