This skill is a set of Markdown instruction files for Claude Code. It does not execute code directly, but it instructs Claude Code to modify settings and environment variables that affect API costs and model behaviour. Security matters.

If you discover a security issue, do not open a public issue.

Instead, email artur@thegeolab.net with:

• Description of the vulnerability
• Steps to reproduce
• Potential impact
• Suggested fix (if you have one)

You will receive a response within 7 days. If the issue is confirmed, a fix will be released and you will be credited (unless you prefer anonymity).

• This skill never references specific API keys or credentials
• Settings recommendations use environment variable names only
• Cost tracking advice directs users to Anthropic's dashboard, not to log credentials locally

• The skill recommends changes to .claude/settings.json and environment variables
• All recommended settings are documented Anthropic features
• No recommendations involve undocumented or unsupported configuration

• Incorrect pricing advice could cause unexpected API spend
• All cost figures should be verified against official Anthropic documentation
• The skill clearly labels estimates vs exact figures

• Routing complex security tasks to Haiku could miss vulnerabilities
• The skill explicitly recommends Opus for security audits
• Users should verify model choice is appropriate for security-sensitive work

1. Verify pricing against Anthropic's official pricing page before setting budgets
2. Set API spend limits in the Anthropic console as a safety net
3. Do not commit settings.json if it contains sensitive environment paths
4. Review model routing before security-sensitive tasks -- do not downgrade to save money on security audits
5. Monitor API usage through the official dashboard, not through local logs that could be tampered with