S-TIER technique
Claude Code simple mode — ablating system prompts makes the model smarter
100k ft — What is this?
Claude Code simple mode — ablating system prompts makes the model smarter
50k ft — Why care?
Boris Cherny at YC Startup School: Claude Code accepts --system-prompt for a custom system prompt, and an undocumented simple mode (env var CLAUDE_CODE_SIMPLE=1) deletes all system and tool prompts. Anthropic uses it as an ablation and finds the model is slightly more intelligent without them, though product prompts still help it behave as users expect.
10k ft — Key takeaways
• Ablate your own agent prompts to see which instructions actually help.
• Over-instruction can cap model capability; keep prompts minimal.
• Try simple mode for raw-capability tasks where product guardrails aren't needed.
5k ft — Tools & techniques
Anthropic, Claude, Claude Code
AI agents, custom system prompts, prompt ablation, prompt engineering, workflow automation
1k ft — Deep-dive / try it
Claude Code creator Boris Cherny says the model can actually become slightly more intelligent when its system prompts are removed. Anthropic discovered this through ablation testing, where the team strips away instructions and tool prompts to measure what is genuinely improving the model. The surprising result was that more prompting didn’t always produce more intelligence. In some cases, the additional instructions were constraining what the model could do.
For founders and developers building with AI, this is a useful reminder that prompting is part of the product architecture. Every instruction added to an agent can shape its behavior, narrow its decision-making, and influence how well it handles problems outside the exact workflow you designed. The goal is to give the model enough con…
Transcript excerpt
So when you run quad code, you can just do, like, dash dash system prompt, and set whatever system prompt you want. And another thing that you can try is simple mode. So this is actually this kind of undocumented feature. If you do quad code simple equals one, like this environment variable, and then you run quad, it'll delete all the system prompts, including from the tools. And we actually use this as a sort of ablation to figure out, is the prompt useful? And what's interesting is that the model is actually a little bit more intelligent without these prompts. That's something that we've been finding. But when you use quad code as a product, you do actually want some of these prompts because it helps you use the product and it helps the product behave and the model behave in the way that you would want when you're using it as a person.