Security
Security Model
How skills are sandboxed and kept secure.
FindSkills takes security seriously. Skills are read-only prompt files — they cannot execute code or access your system directly.
What Skills Can Do
- Provide instructions and context to your AI agent
- Suggest code (which you review before running)
- Reference external documentation (read-only)
What Skills Cannot Do
- Execute code on your machine
- Access your file system directly
- Make network requests
- Store or transmit your data
i
All skills published to the registry are scanned for prompt injection attacks and harmful instructions before approval.