PAI is a PUBLIC version of the personal PAI_DIRECTORY infrastructure
This repository is PUBLIC and visible to everyone on the internet. It's a sanitized, public instance of the personal PAI_DIRECTORY infrastructure. When moving functionality from PAI_DIRECTORY to PAI:
- Personal API keys or tokens
- Private email addresses or phone numbers
- Financial account information
- Health or medical data
- Personal context files
- Business-specific information
- Client or customer data
- Internal URLs or endpoints
- Security credentials
- Personal file paths beyond ${PAI_DIR}
- Generic command structures
- Public documentation
- Example configurations (with placeholder values)
- Open-source integrations
- General-purpose tools
- Public API documentation
- Audit all changes - Review every file being committed
- Search for sensitive data - grep for emails, keys, tokens
- Check context files - Ensure no personal context is included
- Verify paths - All paths should use ${PAI_DIR}, not personal directories
- Test with fresh install - Ensure it works without your personal setup
When copying from PAI_DIRECTORY to PAI:
- Remove all API keys (replace with placeholders)
- Remove personal information
- Replace specific paths with ${PAI_DIR}
- Remove business-specific context
- Sanitize example data
- Update documentation to be generic
- Test in clean environment
- Immediately remove from GitHub
- Revoke any exposed API keys
- Change any exposed passwords
- Use
git filter-branchor BFG to remove from history - Force push cleaned history
- Audit for any data that may have been scraped
- Keep PAI_DIRECTORY private and local
- PAI should be the generic, public template
- Use environment variables for all sensitive config
- Document what needs to be configured by users
- Provide example env-example files, never real .env
Remember: PAI is meant to help everyone build their own personal AI infrastructure. Keep it clean, generic, and safe for public consumption.
When in doubt, DON'T include it in PAI.