XAI released Grok 4.3, introducing account — level Grok Skills and an upgraded Responses API that together make user workflows persistent across web, iOS, and Android and expand how models call external tools. The changes matter because they keep context, preferences, and documents available across conversations and shift some execution from client apps to xAI’s server — side infrastructure, simplifying multi — step automation for users and developers alike.
Grok Skills let users define persistent custom expertise and workflows once and have those instructions applied automatically across the platform and mobile apps. Skills can be written in natural language or created by uploading files; once saved, Grok applies the defined workflows, preferences, and document — handling routines in subsequent chats so users don’t have to repeat setup steps each session.
Built — in Grok Skill capabilities include full Word generation and editing that preserves headings, tables, and styles; PowerPoint — style slide deck creation with visual hierarchy and speaker notes; Excel spreadsheets that support formulas, charts, and conditional formatting; and PDF creation, merging, splitting, and text extraction. Skills live at the account level, take priority when invoked via slash commands, and can be shared between users for collaboration.
On the developer side, Grok 4.3 extends the Responses API with tool-calling that follows an OpenAI — compatible format while adding native server — side execution for built — in tools. Developers can request built — in types such as web_search, x_search, or code_interpreter to run automatically on xAI infrastructure, or define custom functions using JSON schemas that describe name, description, and parameters.
When the model chooses a tool, it returns structured tool_call objects; client apps then execute the logic locally, append the tool outputs in the next request, and continue the conversation loop. The Responses API now runs parallel tool calls by default, supports up to 128 tools per request, and expands context capacity to a 1 million token window — limits positioned to support multi — step, agentic tasks and complex orchestration without frequent context truncation.
Early community reaction on X mixed practical testing with enthusiasm for workflow automation. Developer Tiago Rama wrote, "Custom skills/workflow automation have been becoming the default in other AI tools, so Grok needed to catch up here." Another developer, William Wallace, shared a Grok Skill tied to GitHub and said, "I have enabled Grok to connect to my Github account to read and commit. I added this context.md file to help maintain context across multiple development conversations."
Compared with OpenAI Skills, Claude Skills, and Vercel Agent Skills, Grok Skills is presented as a reusable workflow and capability layer rather than a fully deployable autonomous agent system. xAI positions Grok’s differentiator as tight integration of reusable instructions with X-platform search, multimodal inputs, and social context to produce lightweight, platform — native workflows teams and developers can adopt without building full agent orchestration stacks.
Sources
Replies (0)
No replies in this topic yet.
