Anthropic Cut 80% of Claude’s Prompt, Young Coders Vanish, Microsoft Snubs OpenAI
Anthropic Cuts the Claude Code System Prompt
Anthropic made a bold move with its Claude Code system prompt. The company cut more than 80% of it. This applies to its newest models, including Claude Opus 5 and Claude Fable 5. However, coding evaluation scores did not drop. In fact, that is the surprising part.
For years, engineers stacked rule after rule into the Claude Code system prompt. Some rules said never write code comments. Others banned planning documents entirely. Anthropic now says many of these rules were often wrong for specific situations. As a result, the models can finally be trusted to decide on their own.
Progressive Disclosure Replaces Rigid Rules
Anthropic is shifting toward something it calls progressive disclosure. Because of this shift, context loads only when an agent actually needs it. Tools like ToolSearch help with this. In addition, dedicated skills now handle verification and code review separately. Anthropic also shipped a new command called claude doctor. So developers run /doctor inside Claude Code to audit their own CLAUDE.md files. It flags bloat, outdated rules, and conflicting instructions automatically.
Futurum Group’s Mitch Ashley summed it up well. He said Anthropic cutting 80% of that prompt shows the models can finally be trusted to decide. Therefore, you can read Anthropic’s full breakdown of the change on Techstrong.ai.
A Bigger Shift Across the Industry
This story connects to two other major threads today. Meanwhile, US software developer employment for ages 22-25 has fallen more than 20% since November 2022. That is the month ChatGPT launched. A Stanford Digital Economy Lab analysis of ADP payroll data found a 16% relative employment decline for young coders in AI-exposed roles. For example, AI coding agents like Claude Code and OpenAI Codex now absorb much of the routine work junior developers used to learn on. Read more on our Techstrong Gang show page.
Microsoft also made news today. The company launched MAI-Image-2.5-Pro and MAI-Voice-2-Flash into public preview. These models now power Bing, PowerPoint, OneDrive, Dynamics 365, Excel, GitHub Copilot, and Azure. As a result, this signals a real shift away from heavy OpenAI reliance. It lands right alongside Anthropic’s own Claude Opus 5 launch. But that model pairs near-frontier performance with roughly half the operating cost of its predecessor.
Today’s Techstrong Gang panel breaks down all three stories. Host Alan Shimel leads the discussion with Mike Vizard, Jon Swartz, Stephen Foskett, and Garima Bajpai. So they dig into what leaner AI prompts, vanishing junior coding jobs, and Microsoft’s AI independence play mean for the industry.
