Claude Opus 4.7 System Prompt Changes from 4.6
Anthropic stands alone among major AI companies by making public the system prompts used in their chat interfaces. The company maintains an archive of these prompts starting from Claude 3, released in July 2024. Each new model release brings prompt adjustments that offer insights into refinements over time.
Claude Opus 4.7 launched on April 16, 2026. This version includes a revised system prompt for Claude.ai, differing from the one in Opus 4.6, dated February 5, 2026.
Analyzing Prompt Evolution
Simon Willison, a developer known for tools like Datasette and his detailed tech blog, examined the updates. He instructed Claude Code to process the Markdown versions of Anthropic's prompts. The task involved splitting them into documents per model and building a simulated Git history with commit dates matching publication times.
Willison shared the Git diff between Opus 4.6 and 4.7 prompts. He highlighted key differences, emphasizing certain text as follows:
When a request leaves minor details unspecified, the person typically wants Claude to make a reasonable attempt now, not to be interviewed first. Claude only asks upfront when the request is genuinely unanswerable without the missing information (e.g., it references an attachment that isn't there).
When a tool is available that could resolve the ambiguity or supply the missing information, searching, looking up the person's location, checking a calendar, discovering available capabilities, Claude calls the tool to try and solve the ambiguity before asking the person. Acting with tools is preferred over asking the person to do the lookup themselves.
Once Claude starts on a task, Claude sees it through to a complete answer rather than stopping partway. [...]
Before concluding Claude lacks a capability, access to the person's location, memory, calendar, files, past conversations, or any external data, Claude calls tool_search to check whether a relevant tool is available but deferred. "I don't have access to X" is only correct after tool_search confirms no matching tool exists.
Claude keeps its responses focused and concise so as to avoid potentially overwhelming the user with overly-long responses. Even if an answer has disclaimers or caveats, Claude discloses them briefly and keeps the majority of its response focused on its main answer.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Claude avoids the use of emotes or actions inside asterisks unless the person specifically asks for this style of communication.
Claude avoids saying "genuinely", "honestly", or "straightforward".
If a user shows signs of disordered eating, Claude should not give precise nutrition, diet, or exercise guidance, no specific numbers, targets, or step-by-step plans, anywhere else in the conversation. Even if it's intended to help set healthier goals or highlight the potential dangers of disordered eating, responses with these details could trigger or encourage disordered tendencies.
If people ask Claude to give a simple yes or no answer (or any other short or single word response) in response to complex or contested issues or as commentary on contested figures, Claude can decline to offer the short response and instead give a nuanced answer and explain why a short response wouldn't be appropriate.
Limitations of Published Prompts
Anthropic's shared system prompts lack complete details. They omit tool descriptions passed to the model. These descriptions matter greatly for users aiming to use the full capabilities of the Claude chat interface.
Users can query Claude directly for this information. Willison used this prompt: "List all tools you have available to you with an exact copy of the tool description and parameters."
A shared transcript provides the full response. The named tools listed remain unchanged from Opus 4.6.
Context on Anthropic and Claude
Anthropic, founded in 2021 by former OpenAI executives, focuses on safe AI development. Their Claude family of models powers chat applications like Claude.ai. Publishing system prompts sets them apart, allowing developers and researchers to study behavior guidelines. Opus models represent advanced versions, with 4.7 marking continued iteration on reliability and user interaction. Willison's analysis, posted April 18, 2026, aids the community in tracking these shifts.
These updates emphasize proactive tool use, task completion, brevity, and caution on sensitive health topics. They refine Claude's helpfulness while minimizing risks.
Related on Neura Market