How do I get structured data from Claude?

Use the --output-format json flag combined with --json-schema to force Claude to return data matching a specific JSON structure, which can then be parsed programmatically.

Can I use Claude Headless Mode in CI/CD pipelines?

Yes, by using the --permission-mode bypassPermissions flag, you can run Claude in automated, trusted environments like GitHub Actions, GitLab CI, or Jenkins.

What is the best way to parse JSON results from the CLI?

The recommended approach is to pipe the JSON output into a tool like jq. For example: claude -p 'Task' --output-format json | jq -r '.structured_output'.

What is Claude Headless Mode?

It is a non-interactive execution mode triggered by the --print or -p flag that allows the Claude CLI to run tasks and return output without requiring user prompts or manual interaction.

Is conversation history preserved in headless mode?

No, each invocation of claude -p starts a fresh session. This ensures that tasks are independent and do not carry over context from previous executions.

Claude Headless Mode & Automation

Name: Claude Headless Mode & Automation
Author: majesticlabs-dev

bymajesticlabs-dev

•

開発者ツール

Automates non-interactive tasks using Claude CLI through scripting, CI/CD pipelines, and structured JSON output.

概要

This skill empowers developers to integrate Claude's intelligence directly into automated workflows by leveraging the --print (or -p) headless mode. It provides standardized patterns for generating text, JSON, or streaming responses, and introduces robust structured data extraction using JSON schemas. Ideal for CI/CD linting, batch processing of files, and building custom CLI tools, it ensures consistent output and reliable model behavior in programmatic environments where user interaction is not possible.

主な機能

Structured output validation using customizable JSON schemas
18 GitHub stars
Independent task execution with fresh context per invocation
Non-interactive command execution with --print and --p flags
Integration-ready permission modes including bypassPermissions
Native support for JSON and verbose streaming JSON formats

ユースケース

Building custom shell scripts that utilize AI for logic and decision making
Batch processing directories for data extraction or documentation
Automating code reviews and security audits within CI/CD pipelines

概要

主な機能

Structured output validation using customizable JSON schemas
18 GitHub stars
Independent task execution with fresh context per invocation
Non-interactive command execution with --print and --p flags
Integration-ready permission modes including bypassPermissions
Native support for JSON and verbose streaming JSON formats

ユースケース

Building custom shell scripts that utilize AI for logic and decision making
Batch processing directories for data extraction or documentation
Automating code reviews and security audits within CI/CD pipelines