Learning Log

Weekly notes on what I studied, tried, broke, and fixed while working toward AI Red Teaming. Not polished — this is a working log.

Week 4: Insecure Output Handling — From Toy Demo to Real Attack Chains

Building a toy XSS demo to understand why LLM output shouldn’t be trusted downstream, then tracing how this plays out in real CVEs and NVR-style natural language query systems.

Week 1: OWASP LLM Top 10 and Prompt Injection Testing on Llama 3 and Mistral

Testing three prompt injection techniques against two local LLMs (Llama 3 8B and Mistral 7B) — direct override, roleplay jailbreak, and encoding obfuscation — to see how they hold up.

Week 2: RAG Pipeline Architecture Review

Reviewing the two phases of my RAG pipeline — indexing and query — and drawing out the full data flow.

Week 3: RAG Pipeline

Testing whether a RAG pipeline can be hijacked by an instruction hidden inside a retrieved document — and why a task-specific prompt template may accidentally provide some resistance.