Changes in the system prompt between Claude Opus 4.6 and 4.7 | Hacker News

Hacker News users discussed the system prompt differences between Claude Opus 4.6 and 4.7.

The main changes were more frequent clarification requests and a stronger tendency to caution or refuse ambiguous tasks.

Users also noted Anthropic’s more conservative safety stance on malware-related requests.

Some worried the prompt updates may be temporary or later folded into model training.

The debate underscored how prompt-level policy changes can affect developer workflows and safety transparency.