أنثروبيك تطلق قدرات جديدة قوية لبناء وكلاء الذكاء الاصطناعي

Anthropic has just announced a significant upgrade to its API, introducing four new beta features designed to empower developers in building more sophisticated and capable AI agents. These enhancements, which leverage the power of Claude Opus 4 and Sonnet 4, include a code execution tool, an MCP connector, a Files API, and extended prompt caching capabilities. These features aim to reduce the complexity and cost associated with developing advanced AI agents, allowing developers to focus on innovation rather than infrastructure.

Code Execution Tool: Claude Becomes a Data Analyst

The introduction of the code execution tool transforms Claude from a simple code-writing assistant into a fully fledged data analyst. This tool provides Claude with the ability to run Python code within a secure, sandboxed environment, enabling it to perform complex computational tasks and generate insightful data visualizations directly within API calls. Key benefits and use cases include:

• Enhanced Data Analysis: Claude can now load datasets, create exploratory charts, identify patterns, and refine outputs iteratively.
• Financial Modeling: Generate projections, analyze investment portfolios, and calculate financial metrics.
• Scientific Computing: Execute simulations, process experimental data, and analyze research datasets.
• Business Intelligence: Automate report generation, analyze sales data, and create performance dashboards.
• Document Processing: Extract and transform data from various formats, generate formatted reports, and automate document workflows.
• Statistical Analysis: Perform regression analysis, hypothesis testing, and predictive modeling.

Anthropic is offering 50 free hours of usage per day, with additional usage priced at $0.05 per hour per container.

MCP Connector: Seamless Integration with External Systems

The new MCP (Model Context Protocol) connector dramatically simplifies the process of connecting Claude to remote MCP servers. Previously, developers had to build their own client harnesses to manage these connections. Now, the Anthropic API handles all connection management, tool discovery, and error handling automatically.

• Simplified Integration: Connect to any remote MCP server without writing custom client code.
• Automatic Tool Discovery: Claude automatically connects to specified servers, retrieves available tools, and reasons about which tool to call and what arguments to pass.
• Streamlined Workflow: Automates authentication and error handling, returning enhanced responses with integrated data.
• Ecosystem Expansion: Integrates with a growing ecosystem of remote MCP servers, including Zapier and Asana, enabling seamless access to a wide range of third-party tools.

Files API: Efficient Document Management

The Files API streamlines the way developers store and access documents when building with Claude. Instead of repeatedly uploading files with each request, developers can upload documents once and reference them across multiple conversations.

• Simplified Development: Streamlines workflows, especially for applications dealing with large document sets.
• Integration with Code Execution Tool: Allows Claude to access and process uploaded files directly during code execution, producing outputs like charts and graphs.
• Persistent Data Access: Upload a dataset once and have Claude analyze it across multiple sessions without re-uploading.

Extended Prompt Caching: Cost-Effective Context Management

Anthropic is now offering an extended prompt caching option, allowing developers to choose between the standard 5-minute TTL (Time To Live) and a longer 1-hour TTL. This extended caching can significantly reduce costs and latency, particularly for long-running agent workflows.

• Reduced Costs: Reduce expenses by up to 90% for long prompts.
• Lower Latency: Experience latency reductions of up to 85% for long prompts.
• Enhanced Context Maintenance: Build agents that maintain context over extended periods, ideal for multi-step workflows, complex document analysis, and coordination with other systems.

In conclusion, these new capabilities represent a significant step forward in the development of AI agents on the Anthropic API. By simplifying integration with external systems, enabling powerful data analysis, and offering cost-effective context management, Anthropic is empowering developers to build more sophisticated and practical AI solutions. All features are currently available in public beta.

Source: Anthropic