Anthropic equips Claude Platform with a command-line interface, ant

Anthropic launches ant, a command-line interface enabling developers to execute Claude API endpoints and manage Claude Managed Agents directly from the terminal.

Named `ant`, this new tool makes every Claude API endpoint executable directly from the terminal, from calling the Messages API to creating Claude Managed Agents. Requests are built from typed flags or YAML passed as input rather than manually written JSON, a file is inserted via an `@path` reference, and responses are filtered with `--transform` without third-party tooling. Output can be obtained in JSON, YAML, JSONL, or in an interactive explorer allowing you to collapse and browse the result. The tool also supports versioning API resources in YAML, launching agent sessions, and retrieving their full trace, with every event and tool call remaining accessible. Authentication is done via browser-based OAuth flow or API key, and installation is via Homebrew, curl, or Go. Claude Code can use it natively thanks to a dedicated skill.