Everyone adds AI features to their Laravel app.
Almost nobody threat-models them.
Prompt injection is just SQL injection's younger cousin:
user input reaching a place it was never meant to touch.
Validate LLM input and output like you validate forms.
commit cfc1e5c