Claude Code (Clawdbot)

What it does

This skill wraps the local Claude Code CLI to run headless prompts or interactive tmux sessions. It provides guidance and scripts for running structured JSON output, enabling tool allowlists, and managing interactive slash-command workflows (Spec Kit, OpenSpec). The skill helps agents execute codebase analysis, implement changes, run tests, and recover from TTY/interaction issues.

When to use it

Use when you need to drive the Claude Code (Anthropic) CLI on a host: run repository analysis, plan-then-execute workflows, fix failing tests, or run Spec Kit/OpenSpec flows that require interactive tmux sessions. Prefer interactive mode for slash-command flows that pause for confirmation.

What's included

• Scripts: includes a wrapper script (scripts/claude_code_run.py) to run claude with pseudo-terminal handling.
• References: no separate references directory.
• Instructions: the SKILL.md documents checks (claude --version), patterns for allowlists/permission modes, tips for verification, and operational gotchas (TTY, Vite/ngrok, long-running servers).

Compatible agents

Best for host-based automation agents that can run the claude binary (Claude Code). Useful in CI, developer workstations, and Clawdbot contexts where interactive tmux sessions are available.