EresusSecurity
Reference Database

Language Model Security DB

A top-level reference point for recurring LLM and agent security patterns, integration failure modes, and field-ready research links.

This page connects the reference database to the AI Security Hub, where the practical review paths and Sentinel links live together.

Pattern tracking

Organize repeated issues across prompt injection, tool abuse, unsafe retrieval, identity confusion, and runtime escalation.

Field context

Anchor abstract model-security patterns to practical incident analysis, advisories, and deployment lessons.

Operational use

Give engineering, security, and governance teams one vocabulary for prioritizing model and agent risks.

Reference paths