Agent Stack

An open standard that lets an AI application connect to external tools and data through one client/server interface instead of a bespoke integration per tool.

Folders of instructions, scripts and resources that Claude loads on demand when a task calls for them — packaged, portable expertise rather than a longer prompt.

Distributable bundles that install skills, commands, hooks, MCP servers and agents together, discovered through marketplaces.

Managed integrations in the Claude apps that link Claude to services like Google Drive and Slack, configured by users or administrators rather than in a config file.

Separate Claude Code agents spawned to handle part of a task in their own context window, reporting a result back to the session that launched them.

Multiple full Claude Code sessions orchestrated together, able to message each other, rather than one session delegating to child agents.

The SDK for building custom agents on Claude Code's engine — the same tools, permissions, skills and hooks, with your own orchestration around them.

The mechanism by which Claude calls functions you define, or server-side tools Anthropic hosts, and uses the results to continue its work.

Toolsets that let Claude control a computer or a browser directly — taking screenshots, moving a cursor, typing, and navigating pages.