Skills generated by this tool are markdown files that instruct LLMs to act on your behalf. When loaded by Claude Code, a skill runs with your full user permissions. This means a skill can:

• Read and write files anywhere your user account can access
• Execute shell commands
• Make network requests (if web tools are available)

You are responsible for reviewing generated skills before first use.

Every skill generated by this tool includes:

• A safety notice reminding users to review before execution
• A mandatory constraint prohibiting destructive operations without explicit user confirmation
• No default access to shell or network unless specifically requested

If you find a template or generated pattern that could lead to unintended destructive behavior, data exposure, or permission escalation:

1. Do NOT open a public issue
2. Use GitHub's private vulnerability reporting feature on this repository
3. Include: the template or skill in question, the problematic instruction, and a description of the potential harm

We will acknowledge within 48 hours and provide a fix within 7 days.

• Bugs in Claude Code itself (report to Anthropic)
• Bugs in other LLMs (report to their vendors)
• User-authored skill content that intentionally performs destructive actions