xAI Releases Grok Skills and Updates Tool Calling Responses API

· Source: InfoQ · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

xAI released Grok Skills and updated its Tool Calling Responses API on May 22, 2026, enhancing its Grok 4.3 model. Grok Skills enables users to define persistent custom expertise through natural language or file uploads, which the model retains across all conversations on web, iOS, and Android platforms. These skills support comprehensive document operations, including Word, PowerPoint, Excel, and PDF handling, and can be shared among users. Concurrently, the Responses API now integrates these concepts via an OpenAI-compatible tool calling format, featuring native server-side execution for built-in tools like "web_search" and "code_interpreter". Developers can define custom functions using JSON schemas, with Grok 4.3 supporting parallel tool calls, up to 128 tools per request, and a 1 million token context window for multi-step agentic tasks. This positions Grok Skills as a reusable workflow layer, distinct from broader autonomous agent systems.

Key takeaway

For software engineers building agentic applications or automating complex workflows, Grok Skills and the updated Responses API offer a robust framework. You should explore defining custom skills for persistent, account-level expertise and utilize the OpenAI-compatible tool calling for server-side execution of built-in tools or custom functions. This enables you to build multi-step agents with a 1 million token context window, streamlining document processing and external service integrations.

Key insights

Grok Skills provides persistent, customizable AI expertise and integrated tool calling for automated workflows across platforms.

Principles

Method

Users create skills via natural language or file uploads; developers define custom tools using JSON schemas and process "tool_call" objects client-side.

In practice

Topics

Best for: AI Architect, Machine Learning Engineer, NLP Engineer, AI Engineer, Software Engineer, AI Product Manager

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by InfoQ.