Using LLMs to Set Guardrails for LLMs
Robert Khayat,•AI Safety
How I use Claude/Cursor to configure safety rules that prevent destructive commands.
Coming soon. This post is not written yet. New posts show up in the RSS feed.
Setting up Cursor / Claude Code to deny dangerous scripts like git reset --hard.
Draft Notes:
- The irony of using AI to constrain AI
- Deny lists: git reset, rm -rf, force pushes
- How to configure Claude Code's deny rules
- Cursor settings for safe coding
- Real examples from my setup
Content coming soon...