Claude Code (Official CLI) - Babata Edition 😈

This skill wraps the @anthropic-ai/claude-code tool. It provides a powerful, agentic coding experience directly in your terminal.

🛠️ Prerequisites & Installation

Based on the official docs (https://code.claude.com/docs):

1. Native Install (Recommended):
   • macOS/Linux: curl -fsSL https://claude.ai/install.sh | bash
   • Windows (PowerShell): irm https://claude.ai/install.ps1 | iex
2. Authenticate:

   claude login
   

   Follow the browser flow to connect your Anthropic account.

🚀 Usage Patterns (Optimized for Clawdbot)

Always use pty:true for claude commands to ensure proper interaction and output formatting.

1. One-Shot Agentic Prompt (Recommended)

Use the -p flag for single, non-interactive tasks. This is the primary method for Babata to delegate work.

# General task
bash pty:true workdir:/path/to/project command:"claude -p 'Fix the bug in auth.ts where the token expires too soon'"

# Analysis task
bash pty:true workdir:/path/to/project command:"claude -p 'Read the src/ directory and summarize the architecture'"

2. Sub-Agent Orchestration

Claude Code can spawn multiple agents to work on different parts of a task simultaneously.

bash pty:true command:"claude -p 'Spawn a sub-agent to write tests for the API while I implement the logic'"

3. Git Workflow Automation

Claude Code understands your git history and can automate commits and PRs.

bash pty:true command:"claude -p 'Commit my changes with a descriptive message and push to origin/feature-branch'"

🧠 Memory & Project Context

CLAUDE.md

Create a CLAUDE.md file in your project root to teach Claude about your project's style, build commands, and architecture. This is mandatory for any complex project.

# CLAUDE.md
- Build: `npm run build`
- Test: `npm test`
- Guidelines: Use TypeScript, functional components, and Tailwind.
- Memory: The MT5 bot requires FOK filling mode.

Auto Memory

Claude Code builds "Auto Memory" as it works, saving learnings like build commands and debugging insights across sessions.

🎛️ Key Commands & Flags

Command / Flag	Description
claude	Start interactive session
claude -p "text"	Run one-shot prompt (The "Clawdbot Standard")
/compact	Reduce context window usage (in session)
/cost	Show current session cost (in session)
/init	Initialize a new project with CLAUDE.md

⚠️ Best Practices

1. Workdir Matters: Always specify workdir to the root of the git repository so Claude has full context.
2. Extended Thinking: For complex logic, Claude Code automatically uses extended thinking.
3. Security: Claude Code can run shell commands. Monitor its plan before it executes.
4. Updates: Run npm update -g @anthropic-ai/claude-code regularly or follow the native install update path.