/claude-api

Loads Claude API reference material for the project language (Python, TypeScript, Java, Go, Ruby, C#, PHP, cURL) as well as Managed Agents reference.

[Skill] Covers tool use, streaming, batches, structured outputs, and common pitfalls. Also triggers automatically on imports of 'anthropic'/'@anthropic-ai/sdk'. The 'migrate' subcommand updates existing API code to a newer model (model IDs, thinking configuration, etc.). The 'managed-agents-onboard' subcommand walks interactively through creating a new Managed Agent.

✅ WHEN TO USE IT?

Load reference material before writing your own code against the Claude API

/claude-api — before building a custom streaming handler for the Messages API

Migrate existing API code to a newer model

/claude-api migrate — update model IDs and thinking configuration in existing code

⛔ WHEN NOT TO?

The question is only about operating the Claude Code CLI itself, not your own API code

/claude-api loads API reference material (Python/TS/Java/...), not CLI documentation

Better: use the claude-code-guide agent or the official Claude Code docs for questions about the CLI itself

SOURCES

Claude Code Doku (offiziell) ↗ code.claude.com