claude-code

Configure Anthropic's Claude Code CLI for coding, review, CI/CD workflows.

15|2|Updated May 23, 2026
One-click install
npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill claude-code-vkirill
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: claude-code
Source: https://github.com/VKirill/antigravity-for-claude-code/tree/main/skills/claude-code
Command: npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill claude-code-vkirill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill helps you use Anthropic’s Claude Code CLI to delegate real coding work—without manually juggling hooks, permissions, MCP servers, or CI/headless setup.

Core Features & Use Cases

  • Install, auth, and configuration: Set up @anthropic-ai/claude-code, manage OAuth or ANTHROPIC_API_KEY, and author .claude/settings.json plus CLAUDE.md for project context.
  • Hooks + subagents orchestration: Apply lifecycle hooks (like PreToolUse and PostToolUse) and define narrow, tool-scoped subagents for reviews, planning splits, or repeatable workflows.
  • MCP tool integration + permissions control: Wire MCP servers (stdio/HTTP), constrain tool access with allow/deny matchers and sandbox network rules, and run in plan mode for safety.
  • Headless and CI workflows: Run claude -p ... --output-format json with deterministic exit codes for GitHub Actions and scripts.

Quick Start

Use the claude-code skill to run a read-only pull request review in CI by executing a headless prompt with --permission-mode plan and parsing the resulting JSON .result into a PR comment.

Frequently Asked Questions about claude-code

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run Claude Code in headless mode for CI/CD pipelines?

Run Claude Code in headless mode by executing `claude -p ... --output-format json` with `--permission-mode plan` for deterministic verification in GitHub Actions and scripts.

How do I configure hooks and subagents for automated code review?

Configure hooks by defining `PreToolUse` and `PostToolUse` lifecycle events, and define narrow, tool-scoped subagents in `.claude/settings.json` to orchestrate automated code review workflows.

How do I set up MCP servers and manage permissions for safe tool usage?

Wire MCP servers using stdio or HTTP connections, constrain tool access with allow/deny matchers, and apply sandbox network rules in your settings to ensure safe tool usage.

Can I use Claude Code for read-only pull request reviews in GitHub Actions?

Yes, you can execute a headless prompt with `--permission-mode plan` and parse the resulting JSON `.result` into a PR comment to perform read-only pull request reviews in CI.

What is the best way to delegate coding tasks safely with Claude Code?

Delegate coding safely by applying plan mode, sandbox networking rules, and correct routing/skip logic within your `.claude/settings.json` to guide deterministic verification.

Do I need YAML frontmatter for Claude Code CLI discovery and operation?

Yes, YAML-frontmatter-backed discovery is required for the Claude Code CLI, supporting Pattern 2 reference-driven operation to correctly route and manage related CLI skills.