LLM applications have security problems that ordinary software does not: prompt injection, leaked system prompts, poisoned retrieval documents, output that is used without checks, and models or packages from untrusted sources. These guides explain each risk in plain terms and show defences that hold up in practice.
They begin with how prompt injection works and the OWASP Top 10 for LLM applications, then cover input validation, output verification, hardening model output, RAG security, supply chain trust, red teaming your own system, monitoring with audit logs and a checklist for production. Guardrails for AI agents specifically are covered in the guardrails and security topic hub.
The library lists all guides with filters for topic, level, category and reading time.