Anthropic revamps Console with team prompt sharing and thinking controls

Anthropic just gave its developer console a major overhaul, turning what used to be a standard testing playground into a full-fledged collaborative workspace.

The upgraded Anthropic Console is now available to all users. It introduces team-wide prompt sharing, built-in prompt engineering tools, and native controls for the newly released Claude 3.7 Sonnet.

Shared prompt libraries and one-click code

Building reliable prompts has always been a messy team sport. Engineers, product managers, and QA specialists usually end up copy-pasting raw prompt text across Slack channels and Google Docs, inevitably losing track of versions.

The new Console fixes this by introducing shareable prompts. Teams can now build, refine, and maintain a centralized library of standardized prompts directly inside the platform.

Anthropic also streamlined the iteration loop in the Workbench. You can describe your objective in plain English, and Claude will automatically draft a structured prompt using techniques like chain-of-thought reasoning.

It also includes automated test case generation, side-by-side output evaluation, and a tool to adapt prompts originally written for other AI models. Once you are satisfied with the test results, clicking Get Code generates a production-ready API call instantly.

Taming the extended thinking budget

Alongside the team features, the updated console fully supports Claude 3.7 Sonnet, which can switch between near-instant replies and step-by-step extended thinking.

To keep costs and latency under control, Anthropic added a thinking budget slider. Developers can explicitly set a maximum limit on thinking tokens directly in the console.

You can also flag whether a specific prompt is meant to run with extended thinking turned on, allowing Claude to optimize the prompt structure specifically for deeper reasoning.

Why it matters

We think Anthropic is smartly addressing the biggest operational headache in enterprise AI development: cross-functional collaboration.

In real-world deployments, domain experts know the business logic, but software engineers manage the codebase. Forcing these teams to pass text back and forth through chat apps causes needless friction. Giving them a shared sandbox with automated grading and instant code export makes getting to production much faster.

It also puts pressure on rivals like OpenAI. Developer consoles are shifting from simple playgrounds where you test completions into sticky development platforms. By adding migration tools that adapt competitor prompts for Claude, Anthropic is actively making it easier for teams to switch over.

That said, if you already rely on heavy third-party LLMOps pipelines, the built-in console won't completely replace your testing stack. But for teams currently building on Claude 3.7 Sonnet, running your prompt iteration here is a no-brainer.